BUG: Fix tensorsolve for 0-sized input

`array.reshape(-1, size)` doesn't work with 0 in the dimensions.
The fix is to use explicit shape instead of `-1`. For "non-square"
tensors the `ValueError` would come from the reshape call, while previously
`LinAlgError` appeared from the solve call. To have the same error type
I added a check for squareness before the reshape.

2 files changed, 28 insertions, 1 deletions

diff --git a/numpy/linalg/linalg.py b/numpy/linalg/linalg.py
index 0c27e0631..d831886c0 100644
--- a/numpy/linalg/linalg.py
+++ b/numpy/linalg/linalg.py
@@ -299,7 +299,13 @@ def tensorsolve(a, b, axes=None):
     for k in oldshape:
         prod *= k
 
-    a = a.reshape(-1, prod)
+    if a.size != prod ** 2:
+        raise LinAlgError(
+            "Input arrays must satisfy the requirement \
+            prod(a.shape[b.ndim:]) == prod(a.shape[:b.ndim])"
+        )
+
+    a = a.reshape(prod, prod)
     b = b.ravel()
     res = wrap(solve(a, b))
     res.shape = oldshape
diff --git a/numpy/linalg/tests/test_linalg.py b/numpy/linalg/tests/test_linalg.py
index c1ba84a8e..2462b3996 100644
--- a/numpy/linalg/tests/test_linalg.py
+++ b/numpy/linalg/tests/test_linalg.py
@@ -2103,6 +2103,27 @@ class TestTensorinv:
         assert_allclose(np.tensordot(ainv, b, 1), np.linalg.tensorsolve(a, b))
 
 
+class TestTensorsolve:
+
+    @pytest.mark.parametrize("a, axes", [
+        (np.ones((4, 6, 8, 2)), None),
+        (np.ones((3, 3, 2)), (0, 2)),
+        ])
+    def test_non_square_handling(self, a, axes):
+        with assert_raises(LinAlgError):
+            b = np.ones(a.shape[:2])
+            linalg.tensorsolve(a, b, axes=axes)
+
+    @pytest.mark.parametrize("shape",
+        [(2, 3, 6), (3, 4, 4, 3), (0, 3, 3, 0)],
+    )
+    def test_tensorsolve_result(self, shape):
+        a = np.random.randn(*shape)
+        b = np.ones(a.shape[:2])
+        x = np.linalg.tensorsolve(a, b)
+        assert_allclose(np.tensordot(a, x, axes=len(x.shape)), b)
+
+
 def test_unsupported_commontype():
     # linalg gracefully handles unsupported type
     arr = np.array([[1, -2], [2, 5]], dtype='float16')