def __getitem__(self, key):
# use CSR to implement fancy indexing
if isinstance(key, tuple):
row = key[0]
col = key[1]
if isintlike(row) or isinstance(row, slice):
return self.T[col, row].T
else:
#[[1,2],??] or [[[1],[2]],??]
if isintlike(col) or isinstance(col, slice):
return self.T[col, row].T
else:
row = np.asarray(row, dtype=np.intc)
col = np.asarray(col, dtype=np.intc)
if len(row.shape) == 1:
return self.T[col, row]
elif len(row.shape) == 2:
row = row.reshape(-1)
col = col.reshape(-1, 1)
return self.T[col, row].T
else:
raise NotImplementedError('unsupported indexing')
return self.T[col, row].T
elif isintlike(key) or isinstance(key, slice):
return self.T[:, key].T #[i] or [1:2]
else:
return self.T[:, key].T #[[1,2]]
def __getitem__(self, key):
if isinstance(key, tuple):
row = key[0]
col = key[1]
#TODO implement CSR[ [1,2,3], X ] with sparse matmat
#TODO make use of sorted indices
if isintlike(row) and isintlike(col):
return self._get_single_element(row, col)
else:
major, minor = self._swap((row, col))
if isintlike(major) and isinstance(minor, slice):
minor_shape = self._swap(self.shape)[1]
start, stop, stride = minor.indices(minor_shape)
out_shape = self._swap((1, stop - start))
return self._get_slice(major, start, stop, stride,
out_shape)
elif isinstance(row, slice) or isinstance(col, slice):
return self._get_submatrix(row, col)
else:
raise NotImplementedError
elif isintlike(key):
return self[key, :]
else:
raise IndexError("invalid index")
def __getitem__(self, key):
if isinstance(key, tuple):
row = key[0]
col = key[1]
#TODO implement CSR[ [1,2,3], X ] with sparse matmat
#TODO make use of sorted indices
if isintlike(row) and isintlike(col):
return self._get_single_element(row,col)
else:
major,minor = self._swap((row,col))
if isintlike(major) and isinstance(minor,slice):
minor_shape = self._swap(self.shape)[1]
start, stop, stride = minor.indices(minor_shape)
out_shape = self._swap( (1, stop-start) )
return self._get_slice( major, start, stop, stride, out_shape)
elif isinstance( row, slice) or isinstance(col, slice):
return self._get_submatrix( row, col )
else:
raise NotImplementedError
elif isintlike(key):
return self[key, :]
else:
raise IndexError("invalid index")
def __getitem__(self, key):
# use CSR to implement fancy indexing
if isinstance(key, tuple):
row = key[0]
col = key[1]
if isintlike(row) or isinstance(row, slice):
return self.T[col, row].T
else:
# [[1,2],??] or [[[1],[2]],??]
if isintlike(col) or isinstance(col, slice):
return self.T[col, row].T
else:
row = np.asarray(row, dtype=np.intc)
col = np.asarray(col, dtype=np.intc)
if len(row.shape) == 1:
return self.T[col, row]
elif len(row.shape) == 2:
row = row.reshape(-1)
col = col.reshape(-1, 1)
return self.T[col, row].T
else:
raise NotImplementedError("unsupported indexing")
return self.T[col, row].T
elif isintlike(key) or isinstance(key, slice):
return self.T[:, key].T # [i] or [1:2]
else:
return self.T[:, key].T # [[1,2]]
def get(self, key, default=0.):
"""This overrides the dict.get method, providing type checking
but otherwise equivalent functionality.
"""
try:
i, j = key
assert isintlike(i) and isintlike(j)
except (AssertionError, TypeError, ValueError):
raise IndexError('index must be a pair of integers')
if (i < 0 or i >= self.shape[0] or j < 0 or j >= self.shape[1]):
raise IndexError('index out of bounds')
return dict.get(self, key, default)
def get(self, key, default=0.):
"""This overrides the dict.get method, providing type checking
but otherwise equivalent functionality.
"""
try:
i, j = key
assert isintlike(i) and isintlike(j)
except (AssertionError, TypeError, ValueError):
raise IndexError('index must be a pair of integers')
if (i < 0 or i >= self.shape[0] or j < 0 or j >= self.shape[1]):
raise IndexError('index out of bounds')
return dict.get(self, key, default)
def process_slice(sl, num):
if isinstance(sl, slice):
if sl.step not in (1, None):
raise ValueError('slicing with step != 1 not supported')
i0, i1 = sl.start, sl.stop
if i0 is None:
i0 = 0
elif i0 < 0:
i0 = num + i0
if i1 is None:
i1 = num
elif i1 < 0:
i1 = num + i1
return i0, i1
elif isintlike(sl):
if sl < 0:
sl += num
return sl, sl + 1
else:
raise TypeError('expected slice or scalar')
def __pow__(self, other):
if self.shape[0] != self.shape[1]:
raise TypeError('matrix is not square')
if isintlike(other):
other = int(other)
if other < 0:
raise ValueError('exponent must be >= 0')
if other == 0:
from construct import identity
return identity( self.shape[0], dtype=self.dtype )
elif other == 1:
return self.copy()
else:
result = self
for i in range(1,other):
result = result*self
return result
elif isscalarlike(other):
raise ValueError('exponent must be an integer')
elif isspmatrix(other):
warn('Using ** for elementwise multiplication is deprecated.'\
'Use .multiply() instead', DeprecationWarning)
return self.multiply(other)
else:
raise NotImplementedError
def process_slice(sl, num):
if isinstance(sl, slice):
if sl.step not in (1, None):
raise ValueError("slicing with step != 1 not supported")
i0, i1 = sl.start, sl.stop
if i0 is None:
i0 = 0
elif i0 < 0:
i0 = num + i0
if i1 is None:
i1 = num
elif i1 < 0:
i1 = num + i1
return i0, i1
elif isintlike(sl):
if sl < 0:
sl += num
return sl, sl + 1
else:
raise TypeError("expected slice or scalar")
def __pow__(self, other):
if self.shape[0] != self.shape[1]:
raise TypeError('matrix is not square')
if isintlike(other):
other = int(other)
if other < 0:
raise ValueError('exponent must be >= 0')
if other == 0:
from construct import identity
return identity( self.shape[0], dtype=self.dtype )
elif other == 1:
return self.copy()
else:
result = self
for i in range(1,other):
result = result*self
return result
elif isscalarlike(other):
raise ValueError('exponent must be an integer')
elif isspmatrix(other):
warn('Using ** for elementwise multiplication is deprecated.'\
'Use .multiply() instead', DeprecationWarning)
return self.multiply(other)
else:
raise NotImplementedError
def process_slice(sl, num):
if isinstance(sl, slice):
i0, i1 = sl.start, sl.stop
if i0 is None:
i0 = 0
elif i0 < 0:
i0 = num + i0
if i1 is None:
i1 = num
elif i1 < 0:
i1 = num + i1
return i0, i1
elif isintlike(sl):
if sl < 0:
sl += num
return sl, sl + 1
else:
raise TypeError('expected slice or scalar')
def __pow__(self, other):
if self.shape[0] != self.shape[1]:
raise TypeError('matrix is not square')
if isintlike(other):
other = int(other)
if other < 0:
raise ValueError('exponent must be >= 0')
if other == 0:
from construct import identity
return identity( self.shape[0], dtype=self.dtype )
elif other == 1:
return self.copy()
else:
result = self
for i in range(1,other):
result = result*self
return result
elif isscalarlike(other):
raise ValueError('exponent must be an integer')
else:
raise NotImplementedError
def process_slice( sl, num ):
if isinstance( sl, slice ):
i0, i1 = sl.start, sl.stop
if i0 is None:
i0 = 0
elif i0 < 0:
i0 = num + i0
if i1 is None:
i1 = num
elif i1 < 0:
i1 = num + i1
return i0, i1
elif isintlike( sl ):
if sl < 0:
sl += num
return sl, sl + 1
else:
raise TypeError('expected slice or scalar')
def __getitem__(self, key):
def asindices(x):
try:
x = np.asarray(x, dtype=np.intc)
except:
raise IndexError('invalid index')
else:
return x
def check_bounds(indices, N):
max_indx = indices.max()
if max_indx >= N:
raise IndexError('index (%d) out of range' % max_indx)
min_indx = indices.min()
if min_indx < -N:
raise IndexError('index (%d) out of range' % (N + min_indx))
return (min_indx, max_indx)
def extractor(indices, N):
"""Return a sparse matrix P so that P*self implements
slicing of the form self[[1,2,3],:]
"""
indices = asindices(indices)
(min_indx, max_indx) = check_bounds(indices, N)
if min_indx < 0:
indices = indices.copy()
indices[indices < 0] += N
indptr = np.arange(len(indices) + 1, dtype=np.intc)
data = np.ones(len(indices), dtype=self.dtype)
shape = (len(indices), N)
return csr_matrix((data, indices, indptr), shape=shape)
if isinstance(key, tuple):
row = key[0]
col = key[1]
if isintlike(row):
#[1,??]
if isintlike(col):
return self._get_single_element(row, col) #[i,j]
elif isinstance(col, slice):
return self._get_row_slice(row, col) #[i,1:2]
else:
P = extractor(col, self.shape[1]).T #[i,[1,2]]
return self[row, :] * P
elif isinstance(row, slice):
#[1:2,??]
if isintlike(col) or isinstance(col, slice):
return self._get_submatrix(row, col) #[1:2,j]
else:
P = extractor(col, self.shape[1]).T #[1:2,[1,2]]
return self[row, :] * P
else:
#[[1,2],??] or [[[1],[2]],??]
if isintlike(col) or isinstance(col, slice):
P = extractor(row,
self.shape[0]) #[[1,2],j] or [[1,2],1:2]
return (P * self)[:, col]
else:
row = asindices(row)
col = asindices(col)
if len(row.shape) == 1:
if len(row) != len(col): #[[1,2],[1,2]]
raise IndexError(
'number of row and column indices differ')
check_bounds(row, self.shape[0])
check_bounds(col, self.shape[1])
num_samples = len(row)
val = np.empty(num_samples, dtype=self.dtype)
csr_sample_values(self.shape[0], self.shape[1],
self.indptr, self.indices, self.data,
num_samples, row, col, val)
#val = []
#for i,j in zip(row,col):
# val.append(self._get_single_element(i,j))
return np.asmatrix(val)
elif len(row.shape) == 2:
row = np.ravel(row) #[[[1],[2]],[1,2]]
P = extractor(row, self.shape[0])
return (P * self)[:, col]
else:
raise NotImplementedError('unsupported indexing')
elif isintlike(key) or isinstance(key, slice):
return self[key, :] #[i] or [1:2]
else:
return self[asindices(key), :] #[[1,2]]
def __init__(self, arg1, shape=None, dtype=None, copy=False):
_data_matrix.__init__(self)
if isinstance(arg1, tuple):
if isshape(arg1):
M, N = arg1
self.shape = (M,N)
self.row = np.array([], dtype=np.intc)
self.col = np.array([], dtype=np.intc)
self.data = np.array([], getdtype(dtype, default=float))
else:
try:
obj, ij = arg1
except:
raise TypeError('invalid input format')
try:
if len(ij) != 2:
raise TypeError
except TypeError:
raise TypeError('invalid input format')
self.row = np.array(ij[0], copy=copy, dtype=np.intc)
self.col = np.array(ij[1], copy=copy, dtype=np.intc)
self.data = np.array( obj, copy=copy)
if shape is None:
if len(self.row) == 0 or len(self.col) == 0:
raise ValueError('cannot infer dimensions from zero sized index arrays')
M = self.row.max() + 1
N = self.col.max() + 1
self.shape = (M, N)
else:
# Use 2 steps to ensure shape has length 2.
M, N = shape
self.shape = (M, N)
elif arg1 is None:
# Initialize an empty matrix.
if not isinstance(shape, tuple) or not isintlike(shape[0]):
raise TypeError('dimensions not understood')
warn('coo_matrix(None, shape=(M,N)) is deprecated, ' \
'use coo_matrix( (M,N) ) instead', DeprecationWarning)
self.shape = shape
self.data = np.array([], getdtype(dtype, default=float))
self.row = np.array([], dtype=np.intc)
self.col = np.array([], dtype=np.intc)
else:
if isspmatrix(arg1):
if isspmatrix_coo(arg1) and copy:
self.row = arg1.row.copy()
self.col = arg1.col.copy()
self.data = arg1.data.copy()
self.shape = arg1.shape
else:
coo = arg1.tocoo()
self.row = coo.row
self.col = coo.col
self.data = coo.data
self.shape = coo.shape
else:
#dense argument
try:
M = np.atleast_2d(np.asarray(arg1))
except:
raise TypeError('invalid input format')
if np.rank(M) != 2:
raise TypeError('expected rank <= 2 array or matrix')
self.shape = M.shape
self.row,self.col = (M != 0).nonzero()
self.data = M[self.row,self.col]
if dtype is not None:
self.data = self.data.astype(dtype)
self._check()
def __setitem__(self, key, value):
try:
i, j = key
except (ValueError, TypeError):
raise TypeError("index must be a pair of integers or slices")
# First deal with the case where both i and j are integers
if isintlike(i) and isintlike(j):
if i < 0:
i += self.shape[0]
if j < 0:
j += self.shape[1]
if i < 0 or i >= self.shape[0] or j < 0 or j >= self.shape[1]:
raise IndexError("index out of bounds")
if np.isscalar(value):
if value == 0:
if self.has_key((i,j)):
del self[(i,j)]
else:
dict.__setitem__(self, (i,j), self.dtype.type(value))
else:
raise ValueError('setting an array element with a sequence')
else:
# Either i or j is a slice, sequence, or invalid. If i is a slice
# or sequence, unfold it first and call __setitem__ recursively.
if isinstance(i, slice):
# Is there an easier way to do this?
seq = xrange(i.start or 0, i.stop or self.shape[0], i.step or 1)
elif _is_sequence(i):
seq = i
else:
# Make sure i is an integer. (But allow it to be a subclass of int).
if not isintlike(i):
raise TypeError("index must be a pair of integers or slices")
seq = None
if seq is not None:
# First see if 'value' is another dok_matrix of the appropriate
# dimensions
if isinstance(value, dok_matrix):
if value.shape[1] == 1:
for element in seq:
self[element, j] = value[element, 0]
else:
raise NotImplementedError("setting a 2-d slice of"
" a dok_matrix is not yet supported")
elif np.isscalar(value):
for element in seq:
self[element, j] = value
else:
# See if value is a sequence
try:
if len(seq) != len(value):
raise ValueError("index and value ranges must"
" have the same length")
except TypeError:
# Not a sequence
raise TypeError("unsupported type for"
" dok_matrix.__setitem__")
# Value is a sequence
for element, val in izip(seq, value):
self[element, j] = val # don't use dict.__setitem__
# here, since we still want to be able to delete
# 0-valued keys, do type checking on 'val' (e.g. if
# it's a rank-1 dense array), etc.
else:
# Process j
if isinstance(j, slice):
seq = xrange(j.start or 0, j.stop or self.shape[1], j.step or 1)
elif _is_sequence(j):
seq = j
else:
# j is not an integer
raise TypeError("index must be a pair of integers or slices")
# First see if 'value' is another dok_matrix of the appropriate
# dimensions
if isinstance(value, dok_matrix):
if value.shape[0] == 1:
for element in seq:
self[i, element] = value[0, element]
else:
raise NotImplementedError("setting a 2-d slice of"
" a dok_matrix is not yet supported")
elif np.isscalar(value):
for element in seq:
self[i, element] = value
else:
# See if value is a sequence
try:
if len(seq) != len(value):
raise ValueError("index and value ranges must have"
" the same length")
except TypeError:
# Not a sequence
raise TypeError("unsupported type for dok_matrix.__setitem__")
else:
for element, val in izip(seq, value):
self[i, element] = val
def __getitem__(self, key):
"""If key=(i,j) is a pair of integers, return the corresponding
element. If either i or j is a slice or sequence, return a new sparse
matrix with just these elements.
"""
try:
i, j = key
except (ValueError, TypeError):
raise TypeError('index must be a pair of integers or slices')
# Bounds checking
if isintlike(i):
if i < 0:
i += self.shape[0]
if i < 0 or i >= self.shape[0]:
raise IndexError('index out of bounds')
if isintlike(j):
if j < 0:
j += self.shape[1]
if j < 0 or j >= self.shape[1]:
raise IndexError('index out of bounds')
# First deal with the case where both i and j are integers
if isintlike(i) and isintlike(j):
return dict.get(self, (i,j), 0.)
else:
# Either i or j is a slice, sequence, or invalid. If i is a slice
# or sequence, unfold it first and call __getitem__ recursively.
if isinstance(i, slice):
# Is there an easier way to do this?
seq = xrange(i.start or 0, i.stop or self.shape[0], i.step or 1)
elif _is_sequence(i):
seq = i
else:
# Make sure i is an integer. (But allow it to be a subclass of int).
if not isintlike(i):
raise TypeError('index must be a pair of integers or slices')
seq = None
if seq is not None:
# i is a seq
if isintlike(j):
# Create a new matrix of the correct dimensions
first = seq[0]
last = seq[-1]
if first < 0 or first >= self.shape[0] or last < 0 \
or last >= self.shape[0]:
raise IndexError('index out of bounds')
newshape = (last-first+1, 1)
new = dok_matrix(newshape)
# ** This uses linear time in the size m of dimension 0:
# new[0:seq[-1]-seq[0]+1, 0] = \
# [self.get((element, j), 0) for element in seq]
# ** Instead just add the non-zero elements. This uses
# ** linear time in the number of non-zeros:
for (ii, jj) in self.keys():
if jj == j and ii >= first and ii <= last:
dict.__setitem__(new, (ii-first, 0), \
dict.__getitem__(self, (ii,jj)))
else:
###################################
# We should reshape the new matrix here!
###################################
raise NotImplementedError("fancy indexing supported over"
" one axis only")
return new
# Below here, j is a sequence, but i is an integer
if isinstance(j, slice):
# Is there an easier way to do this?
seq = xrange(j.start or 0, j.stop or self.shape[1], j.step or 1)
elif _is_sequence(j):
seq = j
else:
# j is not an integer
raise TypeError("index must be a pair of integers or slices")
# Create a new matrix of the correct dimensions
first = seq[0]
last = seq[-1]
if first < 0 or first >= self.shape[1] or last < 0 \
or last >= self.shape[1]:
raise IndexError("index out of bounds")
newshape = (1, last-first+1)
new = dok_matrix(newshape)
# ** This uses linear time in the size n of dimension 1:
# new[0, 0:seq[-1]-seq[0]+1] = \
# [self.get((i, element), 0) for element in seq]
# ** Instead loop over the non-zero elements. This is slower
# ** if there are many non-zeros
for (ii, jj) in self.keys():
if ii == i and jj >= first and jj <= last:
dict.__setitem__(new, (0, jj-first), \
dict.__getitem__(self, (ii,jj)))
return new
def __init__(self, arg1, shape=None, dtype=None, copy=False):
_data_matrix.__init__(self)
if isinstance(arg1, tuple):
if isshape(arg1):
M, N = arg1
self.shape = (M, N)
self.row = np.array([], dtype=np.intc)
self.col = np.array([], dtype=np.intc)
self.data = np.array([], getdtype(dtype, default=float))
else:
try:
obj, ij = arg1
except:
raise TypeError('invalid input format')
try:
if len(ij) != 2:
raise TypeError
except TypeError:
raise TypeError('invalid input format')
self.row = np.array(ij[0], copy=copy, dtype=np.intc)
self.col = np.array(ij[1], copy=copy, dtype=np.intc)
self.data = np.array(obj, copy=copy)
if shape is None:
if len(self.row) == 0 or len(self.col) == 0:
raise ValueError(
'cannot infer dimensions from zero sized index arrays'
)
M = self.row.max() + 1
N = self.col.max() + 1
self.shape = (M, N)
else:
# Use 2 steps to ensure shape has length 2.
M, N = shape
self.shape = (M, N)
elif arg1 is None:
# Initialize an empty matrix.
if not isinstance(shape, tuple) or not isintlike(shape[0]):
raise TypeError('dimensions not understood')
warn('coo_matrix(None, shape=(M,N)) is deprecated, ' \
'use coo_matrix( (M,N) ) instead', DeprecationWarning)
self.shape = shape
self.data = np.array([], getdtype(dtype, default=float))
self.row = np.array([], dtype=np.intc)
self.col = np.array([], dtype=np.intc)
else:
if isspmatrix(arg1):
if isspmatrix_coo(arg1) and copy:
self.row = arg1.row.copy()
self.col = arg1.col.copy()
self.data = arg1.data.copy()
self.shape = arg1.shape
else:
coo = arg1.tocoo()
self.row = coo.row
self.col = coo.col
self.data = coo.data
self.shape = coo.shape
else:
#dense argument
try:
M = np.atleast_2d(np.asarray(arg1))
except:
raise TypeError('invalid input format')
if np.rank(M) != 2:
raise TypeError('expected rank <= 2 array or matrix')
self.shape = M.shape
self.row, self.col = M.nonzero()
self.data = M[self.row, self.col]
if dtype is not None:
self.data = self.data.astype(dtype)
self._check()
def __setitem__(self, key, value):
try:
i, j = key
except (ValueError, TypeError):
raise TypeError("index must be a pair of integers or slices")
# First deal with the case where both i and j are integers
if isintlike(i) and isintlike(j):
if i < 0:
i += self.shape[0]
if j < 0:
j += self.shape[1]
if i < 0 or i >= self.shape[0] or j < 0 or j >= self.shape[1]:
raise IndexError("index out of bounds")
if np.isscalar(value):
if value == 0:
if self.has_key((i, j)):
del self[(i, j)]
else:
dict.__setitem__(self, (i, j), self.dtype.type(value))
else:
raise ValueError('setting an array element with a sequence')
else:
# Either i or j is a slice, sequence, or invalid. If i is a slice
# or sequence, unfold it first and call __setitem__ recursively.
if isinstance(i, slice):
# Is there an easier way to do this?
seq = xrange(i.start or 0, i.stop or self.shape[0], i.step
or 1)
elif _is_sequence(i):
seq = i
else:
# Make sure i is an integer. (But allow it to be a subclass of int).
if not isintlike(i):
raise TypeError(
"index must be a pair of integers or slices")
seq = None
if seq is not None:
# First see if 'value' is another dok_matrix of the appropriate
# dimensions
if isinstance(value, dok_matrix):
if value.shape[1] == 1:
for element in seq:
self[element, j] = value[element, 0]
else:
raise NotImplementedError(
"setting a 2-d slice of"
" a dok_matrix is not yet supported")
elif np.isscalar(value):
for element in seq:
self[element, j] = value
else:
# See if value is a sequence
try:
if len(seq) != len(value):
raise ValueError("index and value ranges must"
" have the same length")
except TypeError:
# Not a sequence
raise TypeError("unsupported type for"
" dok_matrix.__setitem__")
# Value is a sequence
for element, val in izip(seq, value):
self[element, j] = val # don't use dict.__setitem__
# here, since we still want to be able to delete
# 0-valued keys, do type checking on 'val' (e.g. if
# it's a rank-1 dense array), etc.
else:
# Process j
if isinstance(j, slice):
seq = xrange(j.start or 0, j.stop or self.shape[1], j.step
or 1)
elif _is_sequence(j):
seq = j
else:
# j is not an integer
raise TypeError(
"index must be a pair of integers or slices")
# First see if 'value' is another dok_matrix of the appropriate
# dimensions
if isinstance(value, dok_matrix):
if value.shape[0] == 1:
for element in seq:
self[i, element] = value[0, element]
else:
raise NotImplementedError(
"setting a 2-d slice of"
" a dok_matrix is not yet supported")
elif np.isscalar(value):
for element in seq:
self[i, element] = value
else:
# See if value is a sequence
try:
if len(seq) != len(value):
raise ValueError("index and value ranges must have"
" the same length")
except TypeError:
# Not a sequence
raise TypeError(
"unsupported type for dok_matrix.__setitem__")
else:
for element, val in izip(seq, value):
self[i, element] = val
def __getitem__(self, key):
"""If key=(i,j) is a pair of integers, return the corresponding
element. If either i or j is a slice or sequence, return a new sparse
matrix with just these elements.
"""
try:
i, j = key
except (ValueError, TypeError):
raise TypeError('index must be a pair of integers or slices')
# Bounds checking
if isintlike(i):
if i < 0:
i += self.shape[0]
if i < 0 or i >= self.shape[0]:
raise IndexError('index out of bounds')
if isintlike(j):
if j < 0:
j += self.shape[1]
if j < 0 or j >= self.shape[1]:
raise IndexError('index out of bounds')
# First deal with the case where both i and j are integers
if isintlike(i) and isintlike(j):
return dict.get(self, (i, j), 0.)
else:
# Either i or j is a slice, sequence, or invalid. If i is a slice
# or sequence, unfold it first and call __getitem__ recursively.
if isinstance(i, slice):
# Is there an easier way to do this?
seq = xrange(i.start or 0, i.stop or self.shape[0], i.step
or 1)
elif _is_sequence(i):
seq = i
else:
# Make sure i is an integer. (But allow it to be a subclass of int).
if not isintlike(i):
raise TypeError(
'index must be a pair of integers or slices')
seq = None
if seq is not None:
# i is a seq
if isintlike(j):
# Create a new matrix of the correct dimensions
first = seq[0]
last = seq[-1]
if first < 0 or first >= self.shape[0] or last < 0 \
or last >= self.shape[0]:
raise IndexError('index out of bounds')
newshape = (last - first + 1, 1)
new = dok_matrix(newshape)
# ** This uses linear time in the size m of dimension 0:
# new[0:seq[-1]-seq[0]+1, 0] = \
# [self.get((element, j), 0) for element in seq]
# ** Instead just add the non-zero elements. This uses
# ** linear time in the number of non-zeros:
for (ii, jj) in self.keys():
if jj == j and ii >= first and ii <= last:
dict.__setitem__(new, (ii-first, 0), \
dict.__getitem__(self, (ii,jj)))
else:
###################################
# We should reshape the new matrix here!
###################################
raise NotImplementedError("fancy indexing supported over"
" one axis only")
return new
# Below here, j is a sequence, but i is an integer
if isinstance(j, slice):
# Is there an easier way to do this?
seq = xrange(j.start or 0, j.stop or self.shape[1], j.step
or 1)
elif _is_sequence(j):
seq = j
else:
# j is not an integer
raise TypeError("index must be a pair of integers or slices")
# Create a new matrix of the correct dimensions
first = seq[0]
last = seq[-1]
if first < 0 or first >= self.shape[1] or last < 0 \
or last >= self.shape[1]:
raise IndexError("index out of bounds")
newshape = (1, last - first + 1)
new = dok_matrix(newshape)
# ** This uses linear time in the size n of dimension 1:
# new[0, 0:seq[-1]-seq[0]+1] = \
# [self.get((i, element), 0) for element in seq]
# ** Instead loop over the non-zero elements. This is slower
# ** if there are many non-zeros
for (ii, jj) in self.keys():
if ii == i and jj >= first and jj <= last:
dict.__setitem__(new, (0, jj-first), \
dict.__getitem__(self, (ii,jj)))
return new
def __getitem__(self, key):
def asindices(x):
try:
x = np.asarray(x, dtype=np.intc)
except:
raise IndexError('invalid index')
else:
return x
def check_bounds(indices,N):
max_indx = indices.max()
if max_indx >= N:
raise IndexError('index (%d) out of range' % max_indx)
min_indx = indices.min()
if min_indx < -N:
raise IndexError('index (%d) out of range' % (N + min_indx))
return (min_indx,max_indx)
def extractor(indices,N):
"""Return a sparse matrix P so that P*self implements
slicing of the form self[[1,2,3],:]
"""
indices = asindices(indices)
(min_indx,max_indx) = check_bounds(indices,N)
if min_indx < 0:
indices = indices.copy()
indices[indices < 0] += N
indptr = np.arange(len(indices) + 1, dtype=np.intc)
data = np.ones(len(indices), dtype=self.dtype)
shape = (len(indices),N)
return csr_matrix((data,indices,indptr), shape=shape)
if isinstance(key, tuple):
row = key[0]
col = key[1]
if isintlike(row):
#[1,??]
if isintlike(col):
return self._get_single_element(row, col) #[i,j]
elif isinstance(col, slice):
return self._get_row_slice(row, col) #[i,1:2]
else:
P = extractor(col,self.shape[1]).T #[i,[1,2]]
return self[row,:]*P
elif isinstance(row, slice):
#[1:2,??]
if isintlike(col) or isinstance(col, slice):
return self._get_submatrix(row, col) #[1:2,j]
else:
P = extractor(col,self.shape[1]).T #[1:2,[1,2]]
return self[row,:]*P
else:
#[[1,2],??] or [[[1],[2]],??]
if isintlike(col) or isinstance(col,slice):
P = extractor(row, self.shape[0]) #[[1,2],j] or [[1,2],1:2]
return (P*self)[:,col]
else:
row = asindices(row)
col = asindices(col)
if len(row.shape) == 1:
if len(row) != len(col): #[[1,2],[1,2]]
raise IndexError('number of row and column indices differ')
check_bounds(row, self.shape[0])
check_bounds(col, self.shape[1])
num_samples = len(row)
val = np.empty(num_samples, dtype=self.dtype)
csr_sample_values(self.shape[0], self.shape[1],
self.indptr, self.indices, self.data,
num_samples, row, col, val)
#val = []
#for i,j in zip(row,col):
# val.append(self._get_single_element(i,j))
return np.asmatrix(val)
elif len(row.shape) == 2:
row = np.ravel(row) #[[[1],[2]],[1,2]]
P = extractor(row, self.shape[0])
return (P*self)[:,col]
else:
raise NotImplementedError('unsupported indexing')
elif isintlike(key) or isinstance(key,slice):
return self[key,:] #[i] or [1:2]
else:
return self[asindices(key),:] #[[1,2]]
