MAINT: sparse: reformat str and repr for sparse arrays, correct 1D coords, improve dtype looks

doc/source/tutorial/sparse.rst

@@ -26,10 +26,10 @@ Sparse arrays are a special kind of array where only a few locations in the arra
  [0, 4, 1, 0],
  [0, 0, 5, 0]])
  >>> sparse
- <3x4 sparse array of type '<class 'numpy.int64'>'
-  with 5 stored elements in COOrdinate format>
+ <COOrdinate sparse array of dtype 'int64'
+ with 5 stored elements and shape (3, 4)>
 
-Note that in our dense array, we have five nonzero values. For example, ``2`` is at location ``0,3``, and ``4`` is at location ``1,1``. All of the other values are zero. The sparse array records these five values *explicitly* (see the ``5 stored elements in COOrdinate format``), and then represents all of the remaining zeros as *implicit* values. 
+Note that in our dense array, we have five nonzero values. For example, ``2`` is at location ``0,3``, and ``4`` is at location ``1,1``. All of the other values are zero. The sparse array records these five values *explicitly* (see the ``5 stored elements and shape (3, 4)``), and then represents all of the remaining zeros as *implicit* values. 
 
 Most sparse array methods work in a similar fashion to dense array methods: 
 
@@ -78,8 +78,8 @@ But, other formats, such as the Compressed Sparse Row (CSR) :func:`csr_array()`
 Sometimes, `scipy.sparse` will return a different sparse matrix format than the input sparse matrix format. For example, the dot product of two sparse arrays in COO format will be a CSR format array: 
 
  >>> sparse @ sparse.T
- <3x3 sparse array of type '<class 'numpy.int64'>'
- with 5 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse array of dtype 'int64'
+ with 5 stored elements and shape (3, 3)>
 
 This change occurs because `scipy.sparse` will change the format of input sparse arrays in order to use the most efficient computational method. 
 
@@ -112,8 +112,8 @@ Using these, we can now define a sparse array without building a dense array fir
 
  >>> csr = sp.sparse.csr_array((data, (row, col)))
  >>> csr
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 5 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse array of dtype 'int64'
+ with 5 stored elements and shape (3, 4)>
 
 Different classes have different constructors, but the :func:`scipy.sparse.csr_array`, :func:`scipy.sparse.csc_array`, and :func:`scipy.sparse.coo_array` allow for this style of construction. 
 
@@ -132,8 +132,8 @@ Then, our sparse array will have *six* stored elements, not five:
 
  >>> csr = sp.sparse.csr_array((data, (row, col)))
  >>> csr
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 6 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse array of dtype 'int64'
+ with 6 stored elements and shape (3, 4)>
 
 The "extra" element is our *explicit zero*. The two are still identical when converted back into a dense array, because dense arrays represent *everything* explicitly: 
 
@@ -149,12 +149,12 @@ The "extra" element is our *explicit zero*. The two are still identical when con
 But, for sparse arithmetic, linear algebra, and graph methods, the value at ``2,3`` will be considered an *explicit zero*. To remove this explicit zero, we can use the ``csr.eliminate_zeros()`` method. This operates on the sparse array *in place*, and removes any zero-value stored elements: 
 
  >>> csr
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 6 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse array of dtype 'int64'
+ with 6 stored elements and shape (3, 4)>
  >>> csr.eliminate_zeros()
  >>> csr
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 5 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse array of dtype 'int64'
+ with 5 stored elements and shape (3, 4)>
 
 Before ``csr.eliminate_zeros()``, there were six stored elements. After, there are only five stored elements. 
 
@@ -168,8 +168,8 @@ In this case, we can see that there are *two* ``data`` values that correspond to
 
  >>> dupes = sp.sparse.coo_array((data, (row, col)))
  >>> dupes
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 6 stored elements in COOrdinate format>
+ <COOrdinate sparse array of dtype 'int64'
+ with 6 stored elements and shape (3, 4)>
 
 Note that there are six stored elements in this sparse array, despite only having five unique locations where data occurs. When these arrays are converted back to dense arrays, the duplicate values are summed. So, at location ``1,1``, the dense array will contain the sum of duplicate stored entries, ``1 + 3``: 
 
@@ -182,8 +182,8 @@ To remove duplicate values within the sparse array itself and thus reduce the nu
 
  >>> dupes.sum_duplicates()
  >>> dupes
- <3x4 sparse array of type '<class 'numpy.int64'>'
- with 5 stored elements in COOrdinate format>
+ <COOrdinate sparse array of dtype 'int64'
+ with 5 stored elements and shape (3, 4)>
 
 Now there are only five stored elements in our sparse array, and it is identical to the array we have been working with throughout this guide: 
 

scipy/io/_fast_matrix_market/__init__.py

@@ -332,8 +332,8 @@ def mmread(source):
 
  >>> m = mmread(StringIO(text))
  >>> m
- <5x5 sparse matrix of type '<class 'numpy.float64'>'
- with 7 stored elements in COOrdinate format>
+ <COOrdinate sparse matrix of dtype 'float64'
+  with 7 stored elements and shape (5, 5)>
  >>> m.toarray()
  array([[0., 0., 0., 0., 0.],
  [0., 0., 1., 0., 0.],

scipy/io/_harwell_boeing/hb.py

@@ -499,6 +499,9 @@ def hb_read(path_or_open_file):
  >>> data = csr_matrix(eye(3)) # create a sparse matrix
  >>> hb_write("data.hb", data) # write a hb file
  >>> print(hb_read("data.hb")) # read a hb file
+ <Compressed Sparse Column sparse matrix of dtype 'float64'
+ with 3 stored elements and shape (3, 3)>
+ Coords Values
  (0, 0) 1.0
  (1, 1) 1.0
  (2, 2) 1.0
@@ -550,6 +553,9 @@ def hb_write(path_or_open_file, m, hb_info=None):
  >>> data = csr_matrix(eye(3)) # create a sparse matrix
  >>> hb_write("data.hb", data) # write a hb file
  >>> print(hb_read("data.hb")) # read a hb file
+ <Compressed Sparse Column sparse matrix of dtype 'float64'
+ with 3 stored elements and shape (3, 3)>
+ Coords Values
  (0, 0) 1.0
  (1, 1) 1.0
  (2, 2) 1.0

scipy/sparse/_base.py

@@ -328,10 +328,9 @@ def imag(self):
  def __repr__(self):
  _, format_name = _formats[self.format]
  sparse_cls = 'array' if isinstance(self, sparray) else 'matrix'
- shape_str = 'x'.join(str(x) for x in self.shape)
  return (
- f"<{shape_str} sparse {sparse_cls} of type '{self.dtype.type}'\n"
- f"\twith {self.nnz} stored elements in {format_name} format>"
+ f"<{format_name} sparse {sparse_cls} of dtype '{self.dtype}'\n"
+ f"\twith {self.nnz} stored elements and shape {self.shape}>"
  )
 
  def __str__(self):
@@ -340,18 +339,23 @@ def __str__(self):
  A = self.tocoo()
 
  # helper function, outputs "(i,j) v"
- def tostr(row, col, data):
- triples = zip(list(zip(row, col)), data)
- return '\n'.join([(' {}\t{}'.format(*t)) for t in triples])
+ def tostr(coords, data):
+ pairs = zip(zip(*(c.tolist() for c in coords)), data)
+ return '\n'.join(f' {idx}\t{val}' for idx, val in pairs)
 
+ out = repr(self)
+ if self.nnz == 0:
+ return out
+
+ out += '\n Coords\tValues\n'
  if self.nnz > maxprint:
  half = maxprint // 2
- out = tostr(A.row[:half], A.col[:half], A.data[:half])
+ out += tostr(tuple(c[:half] for c in A.coords), A.data[:half])
  out += "\n :\t:\n"
- half = maxprint - maxprint//2
- out += tostr(A.row[-half:], A.col[-half:], A.data[-half:])
+ half = maxprint - half
+ out += tostr(tuple(c[-half:] for c in A.coords), A.data[-half:])
  else:
- out = tostr(A.row, A.col, A.data)
+ out += tostr(A.coords, A.data)
 
  return out
 

scipy/sparse/_bsr.py

@@ -220,11 +220,10 @@ def _getnnz(self, axis=None):
  def __repr__(self):
  _, fmt = _formats[self.format]
  sparse_cls = 'array' if isinstance(self, sparray) else 'matrix'
- shape_str = 'x'.join(str(x) for x in self.shape)
- blksz = 'x'.join(str(x) for x in self.blocksize)
+ b = 'x'.join(str(x) for x in self.blocksize)
  return (
- f"<{shape_str} sparse {sparse_cls} of type '{self.dtype.type}'\n"
- f"\twith {self.nnz} stored elements (blocksize = {blksz}) in {fmt} format>"
+ f"<{fmt} sparse {sparse_cls} of dtype '{self.dtype}'\n"
+ f"\twith {self.nnz} stored elements (blocksize={b}) and shape {self.shape}>"
  )
 
  def diagonal(self, k=0):

scipy/sparse/_construct.py

@@ -312,11 +312,11 @@ def identity(n, dtype='d', format=None):
  [ 0., 1., 0.],
  [ 0., 0., 1.]])
  >>> sp.sparse.identity(3, dtype='int8', format='dia')
- <3x3 sparse matrix of type '<class 'numpy.int8'>'
-  with 3 stored elements (1 diagonals) in DIAgonal format>
+ <DIAgonal sparse matrix of dtype 'int8'
+ with 3 stored elements (1 diagonals) and shape (3, 3)>
  >>> sp.sparse.eye_array(3, dtype='int8', format='dia')
- <3x3 sparse array of type '<class 'numpy.int8'>'
-  with 3 stored elements (1 diagonals) in DIAgonal format>
+ <DIAgonal sparse array of dtype 'int8'
+ with 3 stored elements (1 diagonals) and shape (3, 3)>
 
  """
  return eye(n, n, dtype=dtype, format=format)
@@ -351,8 +351,8 @@ def eye_array(m, n=None, *, k=0, dtype=float, format=None):
  [ 0., 1., 0.],
  [ 0., 0., 1.]])
  >>> sp.sparse.eye_array(3, dtype=np.int8)
- <3x3 sparse array of type '<class 'numpy.int8'>'
-  with 3 stored elements (1 diagonals) in DIAgonal format>
+ <DIAgonal sparse array of dtype 'int8'
+ with 3 stored elements (1 diagonals) and shape (3, 3)>
 
  """
  # TODO: delete next 15 lines [combine with _eye()] once spmatrix removed
@@ -430,8 +430,8 @@ def eye(m, n=None, k=0, dtype=float, format=None):
  [ 0., 1., 0.],
  [ 0., 0., 1.]])
  >>> sp.sparse.eye(3, dtype=np.int8)
- <3x3 sparse matrix of type '<class 'numpy.int8'>'
- with 3 stored elements (1 diagonals) in DIAgonal format>
+ <DIAgonal sparse matrix of dtype 'int8'
+ with 3 stored elements (1 diagonals) and shape (3, 3)>
 
  """
  return _eye(m, n, k, dtype, format, False)
@@ -1390,8 +1390,8 @@ def rand(m, n, density=0.01, format="coo", dtype=None, random_state=None):
  >>> from scipy.sparse import rand
  >>> matrix = rand(3, 4, density=0.25, format="csr", random_state=42)
  >>> matrix
- <3x4 sparse matrix of type '<class 'numpy.float64'>'
- with 3 stored elements in Compressed Sparse Row format>
+ <Compressed Sparse Row sparse matrix of dtype 'float64'
+  with 3 stored elements and shape (3, 4)>
  >>> matrix.toarray()
  array([[0.05641158, 0. , 0. , 0.65088847], # random
  [0. , 0. , 0. , 0.14286682],

scipy/sparse/_dia.py

@@ -96,11 +96,10 @@ def __init__(self, arg1, shape=None, dtype=None, copy=False):
  def __repr__(self):
  _, fmt = _formats[self.format]
  sparse_cls = 'array' if isinstance(self, sparray) else 'matrix'
- shape_str = 'x'.join(str(x) for x in self.shape)
- ndiag = self.data.shape[0]
+ d = self.data.shape[0]
  return (
- f"<{shape_str} sparse {sparse_cls} of type '{self.dtype.type}'\n"
- f"\twith {self.nnz} stored elements ({ndiag} diagonals) in {fmt} format>"
+ f"<{fmt} sparse {sparse_cls} of dtype '{self.dtype}'\n"
+ f"\twith {self.nnz} stored elements ({d} diagonals) and shape {self.shape}>"
  )
 
  def _data_mask(self):

scipy/sparse/_extract.py

@@ -93,8 +93,8 @@ def tril(A, k=0, format=None):
  [4, 0, 0, 0, 0],
  [0, 0, 0, 0, 0]])
  >>> tril(A, format='csc')
- <3x5 sparse array of type '<class 'numpy.int32'>'
-  with 4 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse array of dtype 'int32'
+ with 4 stored elements and shape (3, 5)>
 
  """
  coo_sparse = coo_array if isinstance(A, sparray) else coo_matrix
@@ -161,8 +161,8 @@ def triu(A, k=0, format=None):
  [4, 5, 0, 6, 7],
  [0, 0, 8, 9, 0]])
  >>> triu(A, format='csc')
- <3x5 sparse array of type '<class 'numpy.int32'>'
-  with 8 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse array of dtype 'int32'
+ with 8 stored elements and shape (3, 5)>
 
  """
  coo_sparse = coo_array if isinstance(A, sparray) else coo_matrix

scipy/sparse/_lil.py

@@ -111,13 +111,6 @@ def count_nonzero(self):
  _getnnz.__doc__ = _spbase._getnnz.__doc__
  count_nonzero.__doc__ = _spbase.count_nonzero.__doc__
 
- def __str__(self):
- val = ''
- for i, row in enumerate(self.rows):
- for pos, j in enumerate(row):
- val += f" {str((i, j))}\t{str(self.data[i][pos])}\n"
- return val[:-1]
-
  def getrowview(self, i):
  """Returns a view of the 'i'th row (without copying).
  """

scipy/sparse/_matrix_io.py

@@ -38,8 +38,8 @@ def save_npz(file, matrix, compressed=True):
  >>> import scipy as sp
  >>> sparse_matrix = sp.sparse.csc_matrix([[0, 0, 3], [4, 0, 0]])
  >>> sparse_matrix
- <2x3 sparse matrix of type '<class 'numpy.int64'>'
- with 2 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse matrix of dtype 'int64'
+  with 2 stored elements and shape (2, 3)>
  >>> sparse_matrix.toarray()
  array([[0, 0, 3],
  [4, 0, 0]], dtype=int64)
@@ -48,8 +48,8 @@ def save_npz(file, matrix, compressed=True):
  >>> sparse_matrix = sp.sparse.load_npz('/tmp/sparse_matrix.npz')
 
  >>> sparse_matrix
- <2x3 sparse matrix of type '<class 'numpy.int64'>'
- with 2 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse matrix of dtype 'int64'
+  with 2 stored elements and shape (2, 3)>
  >>> sparse_matrix.toarray()
  array([[0, 0, 3],
  [4, 0, 0]], dtype=int64)
@@ -109,8 +109,8 @@ def load_npz(file):
  >>> import scipy as sp
  >>> sparse_array = sp.sparse.csc_array([[0, 0, 3], [4, 0, 0]])
  >>> sparse_array
- <2x3 sparse array of type '<class 'numpy.int64'>'
- with 2 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse array of dtype 'int64'
+  with 2 stored elements and shape (2, 3)>
  >>> sparse_array.toarray()
  array([[0, 0, 3],
  [4, 0, 0]], dtype=int64)
@@ -119,8 +119,8 @@ def load_npz(file):
  >>> sparse_array = sp.sparse.load_npz('/tmp/sparse_array.npz')
 
  >>> sparse_array
- <2x3 sparse array of type '<class 'numpy.int64'>'
- with 2 stored elements in Compressed Sparse Column format>
+ <Compressed Sparse Column sparse array of dtype 'int64'
+ with 2 stored elements and shape (2, 3)>
  >>> sparse_array.toarray()
  array([[0, 0, 3],
  [4, 0, 0]], dtype=int64)

scipy/sparse/csgraph/_reordering.pyx

@@ -59,6 +59,9 @@ def reverse_cuthill_mckee(graph, symmetric_mode=False):
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -215,6 +218,9 @@ def structural_rank(graph):
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 8 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 0) 1

scipy/sparse/csgraph/_shortest_path.pyx

@@ -148,6 +148,9 @@ def shortest_path(csgraph, method='auto',
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -294,6 +297,9 @@ def floyd_warshall(csgraph, directed=True,
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -519,6 +525,9 @@ def dijkstra(csgraph, directed=True, indices=None,
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 4 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -1004,6 +1013,9 @@ def bellman_ford(csgraph, directed=True, indices=None,
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -1241,6 +1253,9 @@ def johnson(csgraph, directed=True, indices=None,
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1
@@ -1771,6 +1786,9 @@ def yen(
  ... ]
  >>> graph = csr_matrix(graph)
  >>> print(graph)
+ <Compressed Sparse Row sparse matrix of dtype 'int64'
+ with 5 stored elements and shape (4, 4)>
+ Coords Values
  (0, 1) 1
  (0, 2) 2
  (1, 3) 1