BUG: sparse: Fix summing duplicates for CSR/CSC creation from (data,c…

…oords) (scipy#20687)

* test and then fix duplicates for CSR/CSC creation from (data,coords)

* remove has_canonical_format check when summing duplicates.

scipy/sparse/_compressed.py

@@ -55,6 +55,7 @@ def __init__(self, arg1, shape=None, dtype=None, copy=False):
  coo = self._coo_container(arg1, shape=shape, dtype=dtype)
  arrays = coo._coo_to_compressed(self._swap)
  self.indptr, self.indices, self.data, self._shape = arrays
+ self.sum_duplicates()
  elif len(arg1) == 3:
  # (data, indices, indptr) format
  (data, indices, indptr) = arg1

scipy/sparse/tests/test_base.py

@@ -3851,6 +3851,10 @@ def test_constructor4(self):
  dense = array([[2**63 + 1, 0], [0, 1]], dtype=np.uint64)
  assert_array_equal(dense, csr.toarray())
 
+ # with duplicates (should sum the duplicates)
+ csr = csr_matrix(([1,1,1,1], ([0,2,2,0], [0,1,1,0])))
+ assert csr.nnz == 2
+
  def test_constructor5(self):
  # infer dimensions from arrays
  indptr = array([0,1,3,3])
@@ -4092,6 +4096,10 @@ def test_constructor4(self):
  csc = csc_matrix((data,ij),(4,3))
  assert_array_equal(arange(12).reshape(4, 3), csc.toarray())
 
+ # with duplicates (should sum the duplicates)
+ csc = csc_matrix(([1,1,1,1], ([0,2,2,0], [0,1,1,0])))
+ assert csc.nnz == 2
+
  def test_constructor5(self):
  # infer dimensions from arrays
  indptr = array([0,1,3,3])
@@ -4506,6 +4514,13 @@ def test_todok_duplicates(self):
  dok = coo.todok()
  assert_array_equal(dok.toarray(), coo.toarray())
 
+ def test_tocompressed_duplicates(self):
+ coo = coo_matrix(([1,1,1,1], ([0,2,2,0], [0,1,1,0])))
+ csr = coo.tocsr()
+ assert_equal(csr.nnz + 2, coo.nnz)
+ csc = coo.tocsc()
+ assert_equal(csc.nnz + 2, coo.nnz)
+
  def test_eliminate_zeros(self):
  data = array([1, 0, 0, 0, 2, 0, 3, 0])
  row = array([0, 0, 0, 1, 1, 1, 1, 1])