[Fail on pytest warnings 1/n] Marking strings with invalid escape sequences as raw strings

pytest.ini

@@ -0,0 +1,11 @@
+[pytest]
+# Configure pytest to fail any test that raises an unexpected warning.
+# This will enable us to catch usages of deprecated features in downstream libraries,
+# before they are hard-deprecated.
+# Furthermore, they will catch any test issues which may hide bugs.
+# The format is `action:message_regex:category:module:line`.
+filterwarnings =
+    # Fail builds on any unexpected warnings.
+    # See https://docs.google.com/document/d/1TVMfmhO0vD1MdkVrqRS9t4om2shMTY9nqdqqOopNv0M for more details.
+    error
+    ignore:.*:

python/ray/_private/runtime_env/context.py

@@ -66,7 +66,7 @@ def exec_worker(self, passthrough_args: List[str], language: Language):
         else:
             executable = "exec "
 
-        passthrough_args = [s.replace(" ", "\ ") for s in passthrough_args]
+        passthrough_args = [s.replace(" ", r"\ ") for s in passthrough_args]
         exec_command = " ".join([f"{executable}"] + passthrough_args)
         command_str = " ".join(self.command_prefix + [exec_command])
         # TODO(SongGuyang): We add this env to command for macOS because it doesn't

python/ray/air/util/tensor_extensions/pandas.py

@@ -335,7 +335,7 @@ def name(self) -> str:
 
     @classmethod
     def construct_from_string(cls, string: str):
-        """
+        r"""
         Construct this type from a string.
 
         This is useful mainly for data types that accept parameters.

python/ray/data/grouped_dataset.py

@@ -394,7 +394,7 @@ def count(self) -> Dataset[U]:
     def sum(
         self, on: Union[KeyFn, List[KeyFn]] = None, ignore_nulls: bool = True
     ) -> Dataset[U]:
-        """Compute grouped sum aggregation.
+        r"""Compute grouped sum aggregation.
 
         This is a blocking operation.
 

release/serve_tests/workloads/deployment_graph_wide_ensemble.py

@@ -1,4 +1,4 @@
-"""
+r"""
 Test that focuses on wide fanout of deployment graph
        -> Node_1
       /          \
@@ -56,7 +56,7 @@ def combine(value_refs):
 def test_wide_fanout_deployment_graph(
     fanout_degree, init_delay_secs=0, compute_delay_secs=0
 ):
-    """
+    r"""
     Test that focuses on wide fanout of deployment graph
         -> Node_1
         /          \

release/serve_tests/workloads/serve_handle_wide_ensemble.py

@@ -1,4 +1,4 @@
-"""
+r"""
 This test is parity of
 release/serve_tests/workloads/deployment_graph_wide_ensemble.py
 Instead of using graph api, the test is using pure handle to

rllib/algorithms/algorithm_config.py

@@ -2301,7 +2301,7 @@ def get_multi_agent_setup(
         spaces: Optional[Dict[PolicyID, Tuple[Space, Space]]] = None,
         default_policy_class: Optional[Type[Policy]] = None,
     ) -> Tuple[MultiAgentPolicyConfigDict, Callable[[PolicyID, SampleBatchType], bool]]:
-        """Compiles complete multi-agent config (dict) from the information in `self`.
+        r"""Compiles complete multi-agent config (dict) from the information in `self`.
 
         Infers the observation- and action spaces, the policy classes, and the policy's
         configs. The returned `MultiAgentPolicyConfigDict` is fully unified and strictly

rllib/algorithms/dqn/dqn.py

@@ -51,7 +51,7 @@
 
 
 class DQNConfig(SimpleQConfig):
-    """Defines a configuration class from which a DQN Algorithm can be built.
+    r"""Defines a configuration class from which a DQN Algorithm can be built.
 
     Example:
         >>> from ray.rllib.algorithms.dqn.dqn import DQNConfig

rllib/algorithms/marwil/marwil.py

@@ -178,7 +178,7 @@ def build(
     ) -> "Algorithm":
         if not self._set_off_policy_estimation_methods:
             deprecation_warning(
-                old="MARWIL used to have off_policy_estimation_methods "
+                old=r"MARWIL used to have off_policy_estimation_methods "
                 "is and wis by default. This has"
                 "changed to off_policy_estimation_methods: \{\}."
                 "If you want to use an off-policy estimator, specify it in"

rllib/algorithms/mbmpo/mbmpo.py

@@ -41,7 +41,7 @@
 
 
 class MBMPOConfig(AlgorithmConfig):
-    """Defines a configuration class from which an MBMPO Algorithm can be built.
+    r"""Defines a configuration class from which an MBMPO Algorithm can be built.
 
     Example:
         >>> from ray.rllib.algorithms.mbmpo import MBMPOConfig

rllib/algorithms/r2d2/r2d2.py

@@ -14,7 +14,7 @@
 
 
 class R2D2Config(DQNConfig):
-    """Defines a configuration class from which a R2D2 Algorithm can be built.
+    r"""Defines a configuration class from which a R2D2 Algorithm can be built.
 
     Example:
         >>> from ray.rllib.algorithms.r2d2.r2d2 import R2D2Config