fix lint after new flake8 release added new style constraints (#13047)

Summary:
fix lint after new flake8 release added new style constraints
Pull Request resolved: https://github.com/pytorch/pytorch/pull/13047

Differential Revision: D10527804

Pulled By: soumith

fbshipit-source-id: 6f4d02662570b6339f69117b61037c8394b0bbd8

test/test_jit.py

@@ -3403,7 +3403,7 @@ a")
             if True:
                 x = [1, 2, 3]
             return
-        with self.assertRaisesRegex(RuntimeError, "previously has type Tensor\[\]"):
+        with self.assertRaisesRegex(RuntimeError, r"previously has type Tensor\[\]"):
             self.checkScript(reassign_from_empty_literal, (), optimize=False)
 
         def reassign_from_empty_builtin():
@@ -5967,7 +5967,7 @@ a")
             def f4(a):
                 torch.cat(a)
 
-        with self.assertRaisesRegex(RuntimeError, 'argument \'tensors\' but found int\[\]'):
+        with self.assertRaisesRegex(RuntimeError, r'argument \'tensors\' but found int\[\]'):
             @torch.jit.script
             def f5(a):
                 torch.cat([3])
@@ -6295,7 +6295,7 @@ a")
                     return x
 
     def test_for_range_no_arg(self):
-        with self.assertRaisesRegex(RuntimeError, 'range\(\) expects 1 argument but got 0'):
+        with self.assertRaisesRegex(RuntimeError, r'range\(\) expects 1 argument but got 0'):
             @torch.jit.script
             def range_no_arg(x):
                 for i in range():
@@ -8836,21 +8836,21 @@ class TestCustomOperators(JitTestCase):
     def test_passing_too_many_args(self):
         with self.assertRaisesRegex(
             RuntimeError,
-            "aten::relu\(\) expected at most 1 argument\(s\) but received 2 argument\(s\)"
+            r"aten::relu\(\) expected at most 1 argument\(s\) but received 2 argument\(s\)"
         ):
             torch.ops.aten.relu(1, 2)
 
     def test_passing_too_few_args(self):
         with self.assertRaisesRegex(
             RuntimeError,
-            "aten::relu\(\) is missing value for argument 'self'."
+            r"aten::relu\(\) is missing value for argument 'self'."
         ):
             torch.ops.aten.relu()
 
     def test_passing_one_positional_but_not_the_second(self):
         with self.assertRaisesRegex(
             RuntimeError,
-            "aten::transpose\(\) is missing value for argument 'dim0'."
+            r"aten::transpose\(\) is missing value for argument 'dim0'."
         ):
             torch.ops.aten.transpose(torch.ones(5, 5))
 

test/test_nn.py

@@ -774,8 +774,8 @@ class TestNN(NNTestCase):
         module = nn.Conv2d(in_channels=3, out_channels=33, kernel_size=10, stride=1, bias=True)
         input = torch.randn(1, 3, 1, 1)
         with self.assertRaisesRegex(RuntimeError,
-                                    'Calculated padded input size per channel: \(1 x 1\). ' +
-                                    'Kernel size: \(10 x 10\). Kernel size can\'t be greater than actual input size'):
+                                    r'Calculated padded input size per channel: \(1 x 1\). ' +
+                                    r'Kernel size: \(10 x 10\). Kernel size can\'t be greater than actual input size'):
             module(input)
 
     def test_invalid_conv3d(self):

tools/autograd/utils.py

@@ -32,7 +32,7 @@ IDENT_REGEX = r'(^|\W){}($|\W)'
 # TODO: Use a real parser here; this will get bamboozled
 # by signatures that contain things like std::array<bool, 2> (note the space)
 def split_name_params(prototype):
-    name, params = re.match('(\w+)\((.*)\)', prototype).groups()
+    name, params = re.match(r'(\w+)\((.*)\)', prototype).groups()
     return name, params.split(', ')
 
 

tools/setup_helpers/cuda.py

@@ -47,13 +47,13 @@ def find_cuda_version(cuda_home):
         candidate_names = [os.path.basename(c) for c in candidate_names]
 
     # suppose version is MAJOR.MINOR.PATCH, all numbers
-    version_regex = re.compile('[0-9]+\.[0-9]+\.[0-9]+')
+    version_regex = re.compile(r'[0-9]+\.[0-9]+\.[0-9]+')
     candidates = [c.group() for c in map(version_regex.search, candidate_names) if c]
     if len(candidates) > 0:
         # normally only one will be retrieved, take the first result
         return candidates[0]
     # if no candidates were found, try MAJOR.MINOR
-    version_regex = re.compile('[0-9]+\.[0-9]+')
+    version_regex = re.compile(r'[0-9]+\.[0-9]+')
     candidates = [c.group() for c in map(version_regex.search, candidate_names) if c]
     if len(candidates) > 0:
         return candidates[0]

torch/_torch_docs.py

@@ -17,7 +17,7 @@ def parse_kwargs(desc):
     }
     """
     # Split on exactly 4 spaces after a newline
-    regx = re.compile("\n\s{4}(?!\s)")
+    regx = re.compile(r"\n\s{4}(?!\s)")
     kwargs = [section.strip() for section in regx.split(desc)]
     kwargs = [section for section in kwargs if len(section) > 0]
     return {desc.split(' ')[0]: desc for desc in kwargs}

torch/nn/modules/adaptive.py

@@ -220,7 +220,7 @@ class AdaptiveLogSoftmaxWithLoss(Module):
         return out
 
     def log_prob(self, input):
-        """ Computes log probabilities for all :math:`n\_classes`
+        r""" Computes log probabilities for all :math:`n\_classes`
 
         Args:
             input (Tensor): a minibatch of examples
@@ -240,7 +240,7 @@ class AdaptiveLogSoftmaxWithLoss(Module):
         return self._get_full_log_prob(input, head_output)
 
     def predict(self, input):
-        """ This is equivalent to `self.log_pob(input).argmax(dim=1)`,
+        r""" This is equivalent to `self.log_pob(input).argmax(dim=1)`,
         but is more efficient in some cases.
 
         Args:

torch/optim/adam.py

@@ -4,7 +4,7 @@ from .optimizer import Optimizer
 
 
 class Adam(Optimizer):
-    """Implements Adam algorithm.
+    r"""Implements Adam algorithm.
 
     It has been proposed in `Adam: A Method for Stochastic Optimization`_.
 

torch/optim/sparse_adam.py

@@ -4,7 +4,7 @@ from .optimizer import Optimizer
 
 
 class SparseAdam(Optimizer):
-    """Implements lazy version of Adam algorithm suitable for sparse tensors.
+    r"""Implements lazy version of Adam algorithm suitable for sparse tensors.
 
     In this variant, only moments that show up in the gradient get updated, and
     only those portions of the gradient get applied to the parameters.

torch/utils/collect_env.py

@@ -65,9 +65,9 @@ def run_and_parse_first_match(run_lambda, command, regex):
 
 def get_conda_packages(run_lambda):
     if get_platform() == 'win32':
-        grep_cmd = 'findstr /R "torch soumith"'
+        grep_cmd = r'findstr /R "torch soumith"'
     else:
-        grep_cmd = 'grep "torch\|soumith"'
+        grep_cmd = r'grep "torch\|soumith"'
     out = run_and_read_all(run_lambda, 'conda list | ' + grep_cmd)
     if out is None:
         return out
@@ -91,7 +91,7 @@ def get_nvidia_driver_version(run_lambda):
 
 def get_gpu_info(run_lambda):
     smi = get_nvidia_smi()
-    uuid_regex = re.compile(' \(UUID: .+?\)')
+    uuid_regex = re.compile(r' \(UUID: .+?\)')
     rc, out, _ = run_lambda(smi + ' -L')
     if rc is not 0:
         return None
@@ -190,9 +190,9 @@ def get_pip_packages(run_lambda):
     # People generally have `pip` as `pip` or `pip3`
     def run_with_pip(pip):
         if get_platform() == 'win32':
-            grep_cmd = 'findstr /R "numpy torch"'
+            grep_cmd = r'findstr /R "numpy torch"'
         else:
-            grep_cmd = 'grep "torch\|numpy"'
+            grep_cmd = r'grep "torch\|numpy"'
         return run_and_read_all(run_lambda, pip + ' list --format=legacy | ' + grep_cmd)
 
     if not PY3:

tox.ini

@@ -1,4 +1,4 @@
 [flake8]
 max-line-length = 120
-ignore = E305,E402,E721,E741,F401,F403,F405,F821,F841,F999
+ignore = E305,E402,E721,E741,F401,F403,F405,F821,F841,F999,W503,W504
 exclude = docs/src,venv,third_party,caffe2,scripts,docs/caffe2,tools/amd_build/pyHIPIFY