BUG: `np.resize` negative shape and subclasses edge case fixes

When a negative entry is passed into `np.resize` shape, an error will
now be raised reliably. Previously the `-1` semantics of reshape allowed
edge cases to pass and return incorrect results.
In the corner case of an empty result shape or empty input shape, the
result type will now be preserved.

1 files changed, 18 insertions, 12 deletions

diff --git a/numpy/core/fromnumeric.py b/numpy/core/fromnumeric.py
index f09f2a465..d63e410f6 100644
--- a/numpy/core/fromnumeric.py
+++ b/numpy/core/fromnumeric.py
@@ -1377,10 +1377,17 @@ def resize(a, new_shape):
 
     See Also
     --------
+    np.reshape : Reshape an array without changing the total size.
+    np.pad : Enlarge and pad an array.
+    np.repeat: Repeat elements of an array.
     ndarray.resize : resize an array in-place.
 
     Notes
     -----
+    When the total size of the array does not change `~numpy.reshape` should
+    be used.  In most other cases either indexing (to reduce the size)
+    or padding (to increase the size) may be a more appropriate solution.
+
     Warning: This functionality does **not** consider axes separately,
     i.e. it does not apply interpolation/extrapolation.
     It fills the return array with the required number of elements, taken
@@ -1404,22 +1411,21 @@ def resize(a, new_shape):
     """
     if isinstance(new_shape, (int, nt.integer)):
         new_shape = (new_shape,)
+
     a = ravel(a)
-    Na = len(a)
-    total_size = um.multiply.reduce(new_shape)
-    if Na == 0 or total_size == 0:
-        return mu.zeros(new_shape, a.dtype)
 
-    n_copies = int(total_size / Na)
-    extra = total_size % Na
+    new_size = 1
+    for dim_length in new_shape:
+        new_size *= dim_length
+        if dim_length < 0:
+            raise ValueError('all elements of `new_shape` must be non-negative')
 
-    if extra != 0:
-        n_copies = n_copies + 1
-        extra = Na - extra
+    if a.size == 0 or new_size == 0:
+        # First case must zero fill. The second would have repeats == 0.
+        return np.zeros_like(a, shape=new_shape)
 
-    a = concatenate((a,) * n_copies)
-    if extra > 0:
-        a = a[:-extra]
+    repeats = -(-new_size // a.size)  # ceil division
+    a = concatenate((a,) * repeats)[:new_size]
 
     return reshape(a, new_shape)