Merge pull request #523 from bethgelab/genattacktests

full rewrite of gen_attack_utils tests, fixes #522

Makefile

@@ -14,6 +14,14 @@ testskipslow:
 	pytest --pdb --skipslow --cov=foolbox --cov-append --backend jax
 	pytest --pdb --skipslow --cov=foolbox --cov-append --backend numpy
 
+.PHONY: testskipslowrev
+testskipslowrev:
+	pytest --pdb --skipslow --cov=foolbox --cov-append --backend numpy
+	pytest --pdb --skipslow --cov=foolbox --cov-append --backend jax
+	pytest --pdb --skipslow --cov=foolbox --cov-append --backend tensorflow
+	pytest --pdb --skipslow --cov=foolbox --cov-append --backend pytorch
+	pytest --pdb --skipslow --cov=foolbox
+
 .PHONY: testattacks
 testattacks:
 	pytest --pdb --cov=foolbox.attacks tests/test_attacks.py

tests/test_gen_attack_utils.py

@@ -1,81 +1,42 @@
 import eagerpy as ep
+import numpy as np
+import pytest
+from typing import Any
 
 from foolbox.attacks.gen_attack_utils import rescale_images
 
 
-def test_pytorch_numpy_compatibility() -> None:
-    import numpy as np
-    import torch
+def test_rescale_axis(request: Any, dummy: ep.Tensor) -> None:
+    backend = request.config.option.backend
+    if backend == "numpy":
+        pytest.skip()
 
     x_np = np.random.uniform(0.0, 1.0, size=(16, 3, 64, 64))
-    x_torch = torch.from_numpy(x_np)
-
     x_np_ep = ep.astensor(x_np)
-    x_torch_ep = ep.astensor(x_torch)
-
     x_up_np_ep = rescale_images(x_np_ep, (16, 3, 128, 128), 1)
-    x_up_torch_ep = rescale_images(x_torch_ep, (16, 3, 128, 128), 1)
-
     x_up_np = x_up_np_ep.numpy()
-    x_up_torch = x_up_torch_ep.numpy()
 
-    assert np.allclose(x_up_np, x_up_torch)
+    x = ep.from_numpy(dummy, x_np)
+    x_ep = ep.astensor(x)
+    x_up_ep = rescale_images(x_ep, (16, 3, 128, 128), 1)
+    x_up = x_up_ep.numpy()
 
+    assert np.allclose(x_up_np, x_up)
 
-def test_pytorch_numpy_compatibility_different_axis() -> None:
-    import numpy as np
-    import torch
 
-    x_np = np.random.uniform(0.0, 1.0, size=(16, 64, 64, 3))
-    x_torch = torch.from_numpy(x_np)
+def test_rescale_axis_nhwc(request: Any, dummy: ep.Tensor) -> None:
+    backend = request.config.option.backend
+    if backend == "numpy":
+        pytest.skip()
 
+    x_np = np.random.uniform(0.0, 1.0, size=(16, 64, 64, 3))
     x_np_ep = ep.astensor(x_np)
-    x_torch_ep = ep.astensor(x_torch)
-
     x_up_np_ep = rescale_images(x_np_ep, (16, 128, 128, 3), -1)
-    x_up_torch_ep = rescale_images(x_torch_ep, (16, 128, 128, 3), -1)
-
     x_up_np = x_up_np_ep.numpy()
-    x_up_torch = x_up_torch_ep.numpy()
-
-    assert np.allclose(x_up_np, x_up_torch)
-
-
-def test_pytorch_tensorflow_compatibility() -> None:
-    import numpy as np
-    import torch
-    import tensorflow as tf
-
-    x_np = np.random.uniform(0.0, 1.0, size=(16, 3, 64, 64))
-    x_torch = torch.from_numpy(x_np)
-    x_tf = tf.convert_to_tensor(x_np)
-
-    x_tf_ep = ep.astensor(x_tf)
-    x_torch_ep = ep.astensor(x_torch)
 
-    x_up_tf_ep = rescale_images(x_tf_ep, (16, 3, 128, 128), 1)
-    x_up_torch_ep = rescale_images(x_torch_ep, (16, 3, 128, 128), 1)
-
-    x_up_tf = x_up_tf_ep.numpy()
-    x_up_torch = x_up_torch_ep.numpy()
-
-    assert np.allclose(x_up_tf, x_up_torch)
-
-
-def test_jax_numpy_compatibility() -> None:
-    import numpy as np
-    import jax.numpy as jnp
-
-    x_np = np.random.uniform(0.0, 1.0, size=(16, 3, 64, 64))
-    x_jax = jnp.array(x_np)
-
-    x_np_ep = ep.astensor(x_np)
-    x_jax_ep = ep.astensor(x_jax)
-
-    x_up_np_ep = rescale_images(x_np_ep, (16, 3, 128, 128), 1)
-    x_up_jax_ep = rescale_images(x_jax_ep, (16, 3, 128, 128), 1)
-
-    x_up_np = x_up_np_ep.numpy()
-    x_up_jax = x_up_jax_ep.numpy()
+    x = ep.from_numpy(dummy, x_np)
+    x_ep = ep.astensor(x)
+    x_up_ep = rescale_images(x_ep, (16, 128, 128, 3), -1)
+    x_up = x_up_ep.numpy()
 
-    assert np.allclose(x_up_np, x_up_jax)
+    assert np.allclose(x_up_np, x_up)