Using logging instead of prints

Author: martinRenou

src/sagemaker/base_predictor.py

@@ -14,6 +14,7 @@
 from __future__ import print_function, absolute_import
 
 import abc
+import logging
 from typing import Any, Tuple
 
 from sagemaker.deprecations import (
@@ -54,6 +55,9 @@
 from sagemaker.lineage.context import EndpointContext
 
 
+logger = logging.getLogger(__name__)
+
+
 class PredictorBase(abc.ABC):
     """An object that encapsulates a deployed model."""
 
@@ -452,7 +456,7 @@ def list_monitors(self):
             endpoint_name=self.endpoint_name
         )
         if len(monitoring_schedules_dict["MonitoringScheduleSummaries"]) == 0:
-            print("No monitors found for endpoint. endpoint: {}".format(self.endpoint_name))
+            logger.debug("No monitors found for endpoint. endpoint: %s", self.endpoint_name)
             return []
 
         monitors = []

src/sagemaker/cli/compatibility/v2/files.py

@@ -22,7 +22,8 @@
 
 from sagemaker.cli.compatibility.v2.ast_transformer import ASTTransformer
 
-LOGGER = logging.getLogger(__name__)
+# Setting LOGGER for backward compatibility, in case users import this...
+logger = LOGGER = logging.getLogger(__name__)
 
 
 class FileUpdater(object):
@@ -59,7 +60,7 @@ def _make_output_dirs_if_needed(self):
             os.makedirs(output_dir)
 
         if os.path.exists(self.output_path):
-            LOGGER.warning("Overwriting file %s", self.output_path)
+            logger.warning("Overwriting file %s", self.output_path)
 
 
 class PyFileUpdater(FileUpdater):

src/sagemaker/fw_utils.py

@@ -474,7 +474,7 @@ def tar_and_upload_dir(
         if s3_resource is None:
             s3_resource = session.resource("s3", region_name=session.region_name)
         else:
-            print("Using provided s3_resource")
+            logger.debug("Using provided s3_resource")
 
         s3_resource.Object(bucket, key).upload_file(tar_file, ExtraArgs=extra_args)
     finally:

src/sagemaker/inference_recommender/inference_recommender_mixin.py

@@ -28,7 +28,8 @@
     "mxnet": "MXNET",
 }
 
-LOGGER = logging.getLogger("sagemaker")
+# Setting LOGGER for backward compatibility, in case users import it...
+logger = LOGGER = logging.getLogger("sagemaker")
 
 
 class Phase:
@@ -145,10 +146,10 @@ def right_size(
         )
 
         if endpoint_configurations or traffic_pattern or stopping_conditions or resource_limit:
-            LOGGER.info("Advanced Job parameters were specified. Running Advanced job...")
+            logger.info("Advanced Job parameters were specified. Running Advanced job...")
             job_type = "Advanced"
         else:
-            LOGGER.info("Advanced Job parameters were not specified. Running Default job...")
+            logger.info("Advanced Job parameters were not specified. Running Default job...")
             job_type = "Default"
 
         self._init_sagemaker_session_if_does_not_exist()
@@ -173,7 +174,7 @@ def right_size(
                 vpc_config=self.vpc_config,
                 enable_network_isolation=self.enable_network_isolation(),
             )
-            LOGGER.warning("Attempting to create new model with name %s", self.name)
+            logger.warning("Attempting to create new model with name %s", self.name)
             self.sagemaker_session.create_model(**create_model_args)
 
         ret_name = self.sagemaker_session.create_inference_recommendations_job(
@@ -281,23 +282,23 @@ def _update_params_for_right_size(
         if accelerator_type:
             raise ValueError("accelerator_type is not compatible with right_size().")
         if instance_type or initial_instance_count:
-            LOGGER.warning(
+            logger.warning(
                 "instance_type or initial_instance_count specified."
                 "Overriding right_size() recommendations."
             )
             return None
         if async_inference_config:
-            LOGGER.warning(
+            logger.warning(
                 "async_inference_config is specified. Overriding right_size() recommendations."
             )
             return None
         if serverless_inference_config:
-            LOGGER.warning(
+            logger.warning(
                 "serverless_inference_config is specified. Overriding right_size() recommendations."
             )
             return None
         if explainer_config:
-            LOGGER.warning(
+            logger.warning(
                 "explainer_config is specified. Overriding right_size() recommendations."
             )
             return None
@@ -363,7 +364,7 @@ def _update_params_for_recommendation_id(
         """
 
         if instance_type is not None and initial_instance_count is not None:
-            LOGGER.warning(
+            logger.warning(
                 "Both instance_type and initial_instance_count are specified,"
                 "overriding the recommendation result."
             )

src/sagemaker/local/entities.py

@@ -669,8 +669,10 @@ def start(self, **kwargs):
         execution = _LocalPipelineExecution(execution_id, self.pipeline, **kwargs)
 
         self._executions[execution_id] = execution
-        print(
-            f"Starting execution for pipeline {self.pipeline.name}. Execution ID is {execution_id}"
+        logger.info(
+            "Starting execution for pipeline %s. Execution ID is %s",
+            self.pipeline.name,
+            execution_id,
         )
         self.last_modified_time = datetime.datetime.now().timestamp()
 
@@ -733,31 +735,32 @@ def update_execution_success(self):
         """Mark execution as succeeded."""
         self.status = _LocalExecutionStatus.SUCCEEDED.value
         self.last_modified_time = datetime.datetime.now().timestamp()
-        print(f"Pipeline execution {self.pipeline_execution_name} SUCCEEDED")
+        logger.info("Pipeline execution %s SUCCEEDED", self.pipeline_execution_name)
 
     def update_execution_failure(self, step_name, failure_message):
         """Mark execution as failed."""
         self.status = _LocalExecutionStatus.FAILED.value
         self.failure_reason = f"Step '{step_name}' failed with message: {failure_message}"
         self.last_modified_time = datetime.datetime.now().timestamp()
-        print(
-            f"Pipeline execution {self.pipeline_execution_name} FAILED because step "
-            f"'{step_name}' failed."
+        logger.info(
+            "Pipeline execution %s FAILED because step '%s' failed.",
+            self.pipeline_execution_name,
+            step_name,
         )
 
     def update_step_properties(self, step_name, step_properties):
         """Update pipeline step execution output properties."""
         self.step_execution.get(step_name).update_step_properties(step_properties)
-        print(f"Pipeline step '{step_name}' SUCCEEDED.")
+        logger.info("Pipeline step '%s' SUCCEEDED.", step_name)
 
     def update_step_failure(self, step_name, failure_message):
         """Mark step_name as failed."""
-        print(f"Pipeline step '{step_name}' FAILED. Failure message is: {failure_message}")
+        logger.info("Pipeline step '%s' FAILED. Failure message is: %s", step_name, failure_message)
         self.step_execution.get(step_name).update_step_failure(failure_message)
 
     def mark_step_executing(self, step_name):
         """Update pipelines step's status to EXECUTING and start_time to now."""
-        print(f"Starting pipeline step: '{step_name}'")
+        logger.info("Starting pipeline step: '%s'", step_name)
         self.step_execution.get(step_name).mark_step_executing()
 
     def _initialize_step_execution(self, steps):

src/sagemaker/local/image.py

@@ -230,7 +230,7 @@ def process(
 
         # Print our Job Complete line to have a similar experience to training on SageMaker where
         # you see this line at the end.
-        print("===== Job Complete =====")
+        logger.info("===== Job Complete =====")
 
     def train(self, input_data_config, output_data_config, hyperparameters, environment, job_name):
         """Run a training job locally using docker-compose.
@@ -310,7 +310,7 @@ def train(self, input_data_config, output_data_config, hyperparameters, environm
 
         # Print our Job Complete line to have a similar experience to training on SageMaker where
         # you see this line at the end.
-        print("===== Job Complete =====")
+        logger.info("===== Job Complete =====")
         return artifacts
 
     def serve(self, model_dir, environment):

src/sagemaker/model.py

@@ -73,7 +73,8 @@
     InferenceRecommenderMixin,
 )
 
-LOGGER = logging.getLogger("sagemaker")
+# Setting LOGGER for backward compatibility, in case users import it...
+logger = LOGGER = logging.getLogger("sagemaker")
 
 NEO_ALLOWED_FRAMEWORKS = set(
     ["mxnet", "tensorflow", "keras", "pytorch", "onnx", "xgboost", "tflite"]
@@ -709,7 +710,7 @@ def _upload_code(self, key_prefix: str, repack: bool = False) -> None:
                     script_name=os.path.basename(self.entry_point),
                 )
 
-            LOGGER.info(
+            logger.info(
                 "Repacking model artifact (%s), script artifact "
                 "(%s), and dependencies (%s) "
                 "into single tar.gz file located at %s. "
@@ -1194,13 +1195,13 @@ def compile(
                 self.image_uri = job_status.get("InferenceImage", None)
                 self._is_compiled_model = True
             else:
-                LOGGER.warning(
+                logger.warning(
                     "The instance type %s is not supported for deployment via SageMaker."
                     "Please deploy the model manually.",
                     target_instance_family,
                 )
         else:
-            LOGGER.warning(
+            logger.warning(
                 "Devices described by Target Platform OS, Architecture and Accelerator are not"
                 "supported for deployment via SageMaker. Please deploy the model manually."
             )
@@ -1387,7 +1388,7 @@ def deploy(
             )
 
         if instance_type and instance_type.startswith("ml.inf") and not self._is_compiled_model:
-            LOGGER.warning(
+            logger.warning(
                 "Your model is not compiled. Please compile your model before using Inferentia."
             )