[CODEMOD][caffe2] replace numpy.bool with bool (pytorch#111432)

Test Plan:
numpy.bool is long deprecated and removed starting numpy-1.20.0 [1]. This replaces all references with equivalent `bool` type using the following oneliner:
```
rg -l 'np\.bool' caffe2 | grep '\.py$' | xargs perl -pi -e 's,\bnp\.bool\b,bool,'
```
1. https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

Differential Revision: D50372711

Pull Request resolved: pytorch#111432
Approved by: https://github.com/Skylion007

caffe2/python/core.py

@@ -1645,7 +1645,7 @@ def do_set(operator):
             return do_set(self.GivenTensorInt64Fill)
         elif array.dtype == str:
             return do_set(self.GivenTensorStringFill)
-        elif array.dtype == np.bool:
+        elif array.dtype == bool:
             return do_set(self.GivenTensorBoolFill)
         else:
             return do_set(self.GivenTensorFill)

caffe2/python/hypothesis_test.py

@@ -61,7 +61,7 @@ def _tensor_and_indices(min_dim=1, max_dim=4, dtype=np.float32,
 _NUMPY_TYPE_TO_ENUM = {
     np.float32: core.DataType.FLOAT,
     np.int32: core.DataType.INT32,
-    np.bool: core.DataType.BOOL,
+    bool: core.DataType.BOOL,
     np.uint8: core.DataType.UINT8,
     np.int8: core.DataType.INT8,
     np.uint16: core.DataType.UINT16,
@@ -514,7 +514,7 @@ def test_last_n_windows(self, X, gc, dc):
         import numpy.testing as npt
         npt.assert_almost_equal(output, new_output, decimal=5)
 
-    @given(dtype=st.sampled_from([np.float32, np.float64, np.int32, np.bool]))
+    @given(dtype=st.sampled_from([np.float32, np.float64, np.int32, bool]))
     @settings(deadline=1000)
     def test_print(self, dtype):
         data = np.random.permutation(6).astype(dtype)
@@ -2134,7 +2134,7 @@ def test_cast(self, a, src, dst, use_name, gc, dc):
         # Casting from a float type outside the range of the integral
         # type is UB.
         ftypes = [np.float32, np.float64]
-        if src in ftypes and dst not in ftypes and dst is not np.bool:
+        if src in ftypes and dst not in ftypes and dst is not bool:
             info = np.iinfo(dst)
             a = np.clip(a, info.min, info.max)
 
@@ -2189,7 +2189,7 @@ def ref(data):
         self.assertDeviceChecks(dc, op, [a], [0])
         self.assertReferenceChecks(gc, op, [a], ref)
 
-    @given(data=_dtypes(dtypes=[np.int32, np.int64, np.float32, np.bool]).
+    @given(data=_dtypes(dtypes=[np.int32, np.int64, np.float32, bool]).
            flatmap(lambda dtype: hu.tensor(
                min_dim=1, dtype=dtype, elements=hu.elements_of_type(dtype))),
            has_input=st.booleans(),
@@ -2201,9 +2201,9 @@ def ref(data):
     def test_constant_fill(self, data, has_input, has_extra_shape, extra_shape,
                            gc, dc):
         dtype = data.dtype.type
-        # in opt mode, np.bool is converted into np.bool_
-        if data.dtype == np.dtype(np.bool):
-            dtype = np.bool
+        # in opt mode, bool is converted into np.bool_
+        if data.dtype == np.dtype(bool):
+            dtype = bool
 
         value = data.item(0)
         gt_shape = data.shape
@@ -2236,15 +2236,15 @@ def ref(inputs=None):
         out, = self.assertReferenceChecks(gc, op, inputs, ref)
         self.assertEqual(dtype, out.dtype)
 
-    @given(data=_dtypes(dtypes=[np.int32, np.int64, np.float32, np.bool]).
+    @given(data=_dtypes(dtypes=[np.int32, np.int64, np.float32, bool]).
         flatmap(lambda dtype: hu.tensor(
             min_dim=1, dtype=dtype, elements=hu.elements_of_type(dtype))),
         **hu.gcs)
     @settings(deadline=1000)
     def test_constant_fill_from_tensor(self, data, gc, dc):
         dtype = data.dtype.type
-        if data.dtype == np.dtype(np.bool):
-            dtype = np.bool
+        if data.dtype == np.dtype(bool):
+            dtype = bool
 
         value = np.array([data.item(0)], dtype=dtype)
         inputs = [data, value]

caffe2/python/hypothesis_test_util.py

@@ -150,7 +150,7 @@ def elements_of_type(dtype=np.float32, filter_=None):
         elems = st.integers(min_value=0, max_value=2 ** 31 - 1)
     elif dtype is np.int64:
         elems = st.integers(min_value=0, max_value=2 ** 63 - 1)
-    elif dtype is np.bool:
+    elif dtype is bool:
         elems = st.booleans()
     else:
         raise ValueError("Unexpected dtype without elements provided")

caffe2/python/ideep/dropout_op_test.py

@@ -29,7 +29,7 @@ def test_dropout_is_test(self, X, in_place, ratio, gc, dc):
         # No sense in checking gradients for test phase
 
         def reference_dropout_test(x):
-            return x, np.ones(x.shape, dtype=np.bool)
+            return x, np.ones(x.shape, dtype=bool)
         self.assertReferenceChecks(
             gc, op, [X], reference_dropout_test,
             # The 'mask' output may be uninitialized
@@ -51,7 +51,7 @@ def test_dropout_ratio0(self, X, in_place, output_mask, gc, dc):
         self.assertDeviceChecks(dc, op, [X], [0])
 
         def reference_dropout_ratio0(x):
-            return (x,) if is_test else (x, np.ones(x.shape, dtype=np.bool))
+            return (x,) if is_test else (x, np.ones(x.shape, dtype=bool))
         self.assertReferenceChecks(
             gc, op, [X], reference_dropout_ratio0, outputs_to_check=[0])
 

caffe2/python/layer_model_helper.py

@@ -150,7 +150,7 @@ def _get_global_constant_initializer_op(
                 op_name = 'GivenTensorInt64Fill'
             elif array.dtype == str:
                 op_name = 'GivenTensorStringFill'
-            elif array.dtype == np.bool:
+            elif array.dtype == bool:
                 op_name = 'GivenTensorBoolFill'
             else:
                 op_name = 'GivenTensorFill'

caffe2/python/operator_test/apmeter_test.py

@@ -20,7 +20,7 @@ def calculate_ap(predictions, labels):
         sortind = np.argsort(-scores, kind='mergesort')
         truth = label[sortind]
         precision = np.cumsum(truth) / num_range
-        ap[k] = precision[truth.astype(np.bool)].sum() / max(1, truth.sum())
+        ap[k] = precision[truth.astype(bool)].sum() / max(1, truth.sum())
     return ap
 
 

caffe2/python/operator_test/dropout_op_test.py

@@ -37,7 +37,7 @@ def test_dropout_is_test(self, X, in_place, ratio, engine, gc, dc):
         # No sense in checking gradients for test phase
 
         def reference_dropout_test(x):
-            return x, np.ones(x.shape, dtype=np.bool)
+            return x, np.ones(x.shape, dtype=bool)
         self.assertReferenceChecks(
             gc, op, [X], reference_dropout_test,
             # The 'mask' output may be uninitialized
@@ -69,7 +69,7 @@ def test_dropout_ratio0(self, X, in_place, output_mask, engine, gc, dc):
             self.assertGradientChecks(gc, op, [X], 0, [0])
 
         def reference_dropout_ratio0(x):
-            return (x,) if is_test else (x, np.ones(x.shape, dtype=np.bool))
+            return (x,) if is_test else (x, np.ones(x.shape, dtype=bool))
         self.assertReferenceChecks(
             gc, op, [X], reference_dropout_ratio0,
             # Don't check the mask with cuDNN because it's packed data
@@ -101,7 +101,7 @@ def test_dropout_ratio1(self, X, in_place, output_mask, engine, gc, dc):
             self.assertGradientChecks(gc, op, [X], 0, [0])
 
         def reference_dropout_ratio1(x):
-            return (x,) if is_test else (np.zeros(x.shape, dtype=np.float), np.zeros(x.shape, dtype=np.bool))
+            return (x,) if is_test else (np.zeros(x.shape, dtype=np.float), np.zeros(x.shape, dtype=bool))
         self.assertReferenceChecks(
             gc, op, [X], reference_dropout_ratio1,
             # Don't check the mask with cuDNN because it's packed data

caffe2/python/operator_test/elementwise_ops_test.py

@@ -994,7 +994,7 @@ def reciprocal_op(X):
             gc, op, [X], 0, [0], stepsize=1e-3, threshold=0.05,
             ensure_outputs_are_inferred=True)
 
-    @given(X=hu.tensor(dtype=np.bool), **hu.gcs)
+    @given(X=hu.tensor(dtype=bool), **hu.gcs)
     @settings(deadline=10000)
     def test_not(self, X, gc, dc):
         def not_op(X):

caffe2/python/operator_test/feature_maps_ops_test.py

@@ -27,7 +27,7 @@ def test_merge_dense_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in1_presence",
-            np.array([[True, False, False, True], [False, True, True, False]], dtype=np.bool)
+            np.array([[True, False, False, True], [False, True, True, False]], dtype=bool)
         )
 
         workspace.RunOperatorOnce(op)
@@ -66,7 +66,7 @@ def test_merge_single_scalar_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         # Input 2.
         workspace.FeedBlob(
@@ -75,7 +75,7 @@ def test_merge_single_scalar_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
 
         workspace.RunOperatorOnce(op)
@@ -110,15 +110,15 @@ def test_merge_single_scalar_feature_tensors_gradient(self):
         # Inputs 1, 2 & 3.
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
         workspace.FeedBlob(
             "in3_presence",
-            np.array([False, True], dtype=np.bool)
+            np.array([False, True], dtype=bool)
         )
         # Input 4.
         workspace.FeedBlob(
@@ -158,15 +158,15 @@ def test_merge_single_scalar_feature_tensors_gradient_with_strings(self):
         # Inputs 1, 2 & 3.
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
         workspace.FeedBlob(
             "in3_presence",
-            np.array([False, True], dtype=np.bool)
+            np.array([False, True], dtype=bool)
         )
         # Input 4.
         workspace.FeedBlob(
@@ -214,7 +214,7 @@ def test_merge_single_list_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         # Input 2.
         workspace.FeedBlob(
@@ -227,7 +227,7 @@ def test_merge_single_list_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
 
         workspace.RunOperatorOnce(op)
@@ -280,7 +280,7 @@ def _test_merge_single_list_or_map_feature_tensors_gradient(self, op_name):
         )
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         # Input 2.
         workspace.FeedBlob(
@@ -289,7 +289,7 @@ def _test_merge_single_list_or_map_feature_tensors_gradient(self, op_name):
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
         workspace.FeedBlob(
             "out_values_values_grad",
@@ -336,7 +336,7 @@ def test_merge_single_map_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in1_presence",
-            np.array([True, False], dtype=np.bool)
+            np.array([True, False], dtype=bool)
         )
         # Input 2.
         workspace.FeedBlob(
@@ -353,7 +353,7 @@ def test_merge_single_map_feature_tensors(self):
         )
         workspace.FeedBlob(
             "in2_presence",
-            np.array([True, True], dtype=np.bool)
+            np.array([True, True], dtype=bool)
         )
 
         workspace.RunOperatorOnce(op)

caffe2/python/operator_test/load_save_test.py

@@ -42,7 +42,7 @@ def __init__(self, methodName, db_type='minidb'):
     def load_save(self, src_device_type, src_gpu_id,
                   dst_device_type, dst_gpu_id):
         workspace.ResetWorkspace()
-        dtypes = [np.float16, np.float32, np.float64, np.bool, np.int8,
+        dtypes = [np.float16, np.float32, np.float64, bool, np.int8,
                   np.int16, np.int32, np.int64, np.uint8, np.uint16]
         arrays = [np.random.permutation(6).reshape(2, 3).astype(T)
                   for T in dtypes]
@@ -127,7 +127,7 @@ def _LoadTest(keep_device, device_type, gpu_id, blobs, loadAll):
     def saveFile(
         self, tmp_folder: Path, db_name: str, db_type: str, start_blob_id: int
     ) -> Tuple[str, List[np.ndarray]]:
-        dtypes = [np.float16, np.float32, np.float64, np.bool, np.int8,
+        dtypes = [np.float16, np.float32, np.float64, bool, np.int8,
                   np.int16, np.int32, np.int64, np.uint8, np.uint16]
         arrays = [np.random.permutation(6).reshape(2, 3).astype(T)
                   for T in dtypes]
@@ -153,7 +153,7 @@ def testLoadSave(self):
         self.load_save()
 
     def testRepeatedArgs(self):
-        dtypes = [np.float16, np.float32, np.float64, np.bool, np.int8,
+        dtypes = [np.float16, np.float32, np.float64, bool, np.int8,
                   np.int16, np.int32, np.int64, np.uint8, np.uint16]
         arrays = [np.random.permutation(6).reshape(2, 3).astype(T)
                   for T in dtypes]

caffe2/python/operator_test/onnx_while_test.py

@@ -68,7 +68,7 @@ def test_onnx_while_fibb(
             disable_scopes=disable_scopes,
         )
 
-        condition_arr = np.array(condition).astype(np.bool)
+        condition_arr = np.array(condition).astype(bool)
         max_trip_count_arr = np.array(max_trip_count).astype(np.int64)
         first_init = np.array([1]).astype(np.float32)
         second_init = np.array([1]).astype(np.float32)

caffe2/python/operator_test/pack_ops_test.py

@@ -47,7 +47,7 @@ def pack_segments_ref(lengths, data, max_length=max_length):
                     pad_length = max_length - length
                     presence_arr.append(
                         np.pad(
-                            np.ones((length), dtype=np.bool), ((0, pad_length)),
+                            np.ones((length), dtype=bool), ((0, pad_length)),
                             mode=str("constant")
                         )
                     )
@@ -305,7 +305,7 @@ def test_presence_mask(self, gc, dc):
         presence_mask = workspace.FetchBlob('p')
         expected_presence_mask = np.array(
             [[True, False, False], [True, True, False], [True, True, True]],
-            dtype=np.bool
+            dtype=bool
         )
         self.assertEqual(presence_mask.shape, expected_presence_mask.shape)
         np.testing.assert_array_equal(presence_mask, expected_presence_mask)

caffe2/python/operator_test/shape_inference_test.py

@@ -474,7 +474,7 @@ def testCast(self):
         model = model_helper.ModelHelper(name="test_model")
 
         types = [
-            ('bool', np.bool, caffe2_pb2.TensorProto.BOOL),
+            ('bool', bool, caffe2_pb2.TensorProto.BOOL),
             #('byte', None, caffe2_pb2.TensorProto.BYTE),
             ('int8', np.int8, caffe2_pb2.TensorProto.INT8),
             ('uint8', np.uint8, caffe2_pb2.TensorProto.UINT8),

caffe2/python/schema.py

@@ -1256,7 +1256,7 @@ def InitEmptyRecord(net, schema_or_record, enforce_types=False):
     (np.float16, core.DataType.FLOAT16),
     (np.float32, core.DataType.FLOAT),
     (np.float64, core.DataType.DOUBLE),
-    (np.bool, core.DataType.BOOL),
+    (bool, core.DataType.BOOL),
     (np.int8, core.DataType.INT8),
     (np.int16, core.DataType.INT16),
     (np.int32, core.DataType.INT32),

caffe2/python/sparse_to_dense_mask_test.py

@@ -150,7 +150,7 @@ def test_sparse_to_dense_mask_presence_mask(self):
         expected_output = np.array([[11, -1], [-1, 12]], dtype=np.float)
         expected_presence_mask = np.array(
             [[True, False], [False, True]],
-            dtype=np.bool)
+            dtype=bool)
         self.assertEqual(output.shape, expected_output.shape)
         np.testing.assert_array_equal(output, expected_output)
         self.assertEqual(presence_mask.shape, expected_presence_mask.shape)

caffe2/python/workspace_test.py

@@ -184,7 +184,7 @@ def testFetchFeedBlobTypes(self):
             np.float16,
             np.float32,
             np.float64,
-            np.bool,
+            bool,
             np.int8,
             np.int16,
             np.int32,
@@ -205,20 +205,20 @@ def testFetchFeedBlobTypes(self):
 
     def testFetchFeedBlobBool(self):
         """Special case for bool to ensure coverage of both true and false."""
-        data = np.zeros((2, 3, 4)).astype(np.bool)
+        data = np.zeros((2, 3, 4)).astype(bool)
         data.flat[::2] = True
         self.assertEqual(workspace.FeedBlob("testblob_types", data), True)
         fetched_back = workspace.FetchBlob("testblob_types")
         self.assertEqual(fetched_back.shape, (2, 3, 4))
-        self.assertEqual(fetched_back.dtype, np.bool)
+        self.assertEqual(fetched_back.dtype, bool)
         np.testing.assert_array_equal(fetched_back, data)
 
     def testGetBlobSizeBytes(self):
         for dtype in [
             np.float16,
             np.float32,
             np.float64,
-            np.bool,
+            bool,
             np.int8,
             np.int16,
             np.int32,

torch/_torch_docs.py

@@ -4396,7 +4396,7 @@ def merge_dicts(*dicts):
 It currently accepts :attr:`ndarray` with dtypes of ``numpy.float64``,
 ``numpy.float32``, ``numpy.float16``, ``numpy.complex64``, ``numpy.complex128``,
 ``numpy.int64``, ``numpy.int32``, ``numpy.int16``, ``numpy.int8``, ``numpy.uint8``,
-and ``numpy.bool``.
+and ``bool``.
 
 .. warning::
     Writing to a tensor created from a read-only NumPy array is not supported and will result in undefined behavior.