Merge pull request #671 from lanpa/fix-linting-E275

fix lint: requires whitespace around keywords

tensorboardX/caffe2_graph.py

@@ -292,7 +292,7 @@ def _fill_missing_operator_names(ops):
             name = os.path.join(scope, op.type)
         else:
             name = op.type
-        assert(name)
+        assert name
         op.name = _make_unique_name(seen, name)
 
 
@@ -449,7 +449,7 @@ def _operator_to_node_simp(op, inter_blobs, seen):
             name_list = [name for name in outputs]
             scope = os.path.commonprefix(name_list)
             name = os.path.join(scope, op.type)
-        assert(name)
+        assert name
         op.name = _make_unique_name(seen, name)
         device = _tf_device(op.device_option)
 

tensorboardX/summary.py

@@ -149,7 +149,7 @@ def scalar(name, scalar, display_name="", summary_description=""):
     """
     name = _clean_tag(name)
     scalar = make_np(scalar)
-    assert(scalar.squeeze().ndim == 0), 'scalar should be 0D'
+    assert scalar.squeeze().ndim == 0, 'scalar should be 0D'
     scalar = float(scalar)
     if display_name == "" and summary_description == "":
         return Summary(value=[Summary.Value(tag=name, simple_value=scalar)])
@@ -411,7 +411,7 @@ def audio(tag, tensor, sample_rate=44100):
     if tensor.ndim == 1:  # old API, which expects single channel audio
         tensor = np.expand_dims(tensor, axis=1)
 
-    assert(tensor.ndim == 2), 'Input tensor should be 2 dimensional.'
+    assert tensor.ndim == 2, 'Input tensor should be 2 dimensional.'
     length_frames, num_channels = tensor.shape
     assert num_channels == 1 or num_channels == 2, 'The second dimension should be 1 or 2.'
 

tensorboardX/utils.py

@@ -92,10 +92,10 @@ def make_grid(I, ncols=8):
 
 
 def convert_to_NTCHW(tensor, input_format):
-    assert(len(input_format) == 5), "Only 5D tensor is supported."
-    assert(len(set(input_format)) == len(input_format)), "You can not use the same dimension shordhand twice. \
+    assert len(input_format) == 5, "Only 5D tensor is supported."
+    assert len(set(input_format)) == len(input_format), "You can not use the same dimension shorthand twice. \
         input_format: {}".format(input_format)
-    assert(len(tensor.shape) == len(input_format)), "size of input tensor and input format are different. \
+    assert len(tensor.shape) == len(input_format), "size of input tensor and input format are different. \
         tensor shape: {}, input_format: {}".format(tensor.shape, input_format)
     input_format = input_format.upper()
     index = [input_format.find(c) for c in 'NTCHW']
@@ -105,9 +105,9 @@ def convert_to_NTCHW(tensor, input_format):
 
 def convert_to_HWC(tensor, input_format):  # tensor: numpy array
     import numpy as np
-    assert(len(set(input_format)) == len(input_format)), "You can not use the same dimension shordhand twice. \
+    assert len(set(input_format)) == len(input_format), "You can not use the same dimension shorthand twice. \
         input_format: {}".format(input_format)
-    assert(len(tensor.shape) == len(input_format)), "size of input tensor and input format are different. \
+    assert len(tensor.shape) == len(input_format), "size of input tensor and input format are different. \
         tensor shape: {}, input_format: {}".format(tensor.shape, input_format)
     input_format = input_format.upper()