Coverage for /usr/lib/python3/dist-packages/scipy/sparse/csgraph/_validation.py : 21%

from __future__ import division, print_function, absolute_import 

import numpy as np 

from scipy.sparse import csr_matrix, isspmatrix, isspmatrix_csc 

from ._tools import csgraph_to_dense, csgraph_from_dense,\ 

csgraph_masked_from_dense, csgraph_from_masked 

DTYPE = np.float64 

def validate_graph(csgraph, directed, dtype=DTYPE, 

csr_output=True, dense_output=True, 

copy_if_dense=False, copy_if_sparse=False, 

null_value_in=0, null_value_out=np.inf, 

infinity_null=True, nan_null=True): 

"""Routine for validation and conversion of csgraph inputs""" 

if not (csr_output or dense_output): 

raise ValueError("Internal: dense or csr output must be true") 

# if undirected and csc storage, then transposing in-place 

# is quicker than later converting to csr. 

if (not directed) and isspmatrix_csc(csgraph): 

csgraph = csgraph.T 

if isspmatrix(csgraph): 

if csr_output: 

csgraph = csr_matrix(csgraph, dtype=DTYPE, copy=copy_if_sparse) 

else: 

csgraph = csgraph_to_dense(csgraph, null_value=null_value_out) 

elif np.ma.isMaskedArray(csgraph): 

if dense_output: 

mask = csgraph.mask 

csgraph = np.array(csgraph.data, dtype=DTYPE, copy=copy_if_dense) 

csgraph[mask] = null_value_out 

else: 

csgraph = csgraph_from_masked(csgraph) 

else: 

if dense_output: 

csgraph = csgraph_masked_from_dense(csgraph, 

copy=copy_if_dense, 

null_value=null_value_in, 

nan_null=nan_null, 

infinity_null=infinity_null) 

mask = csgraph.mask 

csgraph = np.asarray(csgraph.data, dtype=DTYPE) 

csgraph[mask] = null_value_out 

else: 

csgraph = csgraph_from_dense(csgraph, null_value=null_value_in, 

infinity_null=infinity_null, 

nan_null=nan_null) 

if csgraph.ndim != 2: 

raise ValueError("compressed-sparse graph must be two dimensional") 

if csgraph.shape[0] != csgraph.shape[1]: 

raise ValueError("compressed-sparse graph must be shape (N, N)") 

return csgraph