fix: use pathlib.PurePosixPath for S3 URLs and Unix paths

This is the Python 3.4+ fix that partially addressed by:

- e86da62
- edd10aa
- 06a00d4
- d2f3984
- 1463743

Author: laurenyu

src/sagemaker/local/utils.py

@@ -14,6 +14,7 @@
 from __future__ import absolute_import
 
 import os
+import pathlib
 import shutil
 
 from distutils.dir_util import copy_tree
@@ -84,8 +85,7 @@ def _create_s3_prefix(path, job_name):
     Returns:
         str: an S3 prefix of the form ``"path/job_name"``
     """
-    path = path.strip("/")
-    return job_name if path == "" else "/".join((path, job_name))
+    return job_name if path == "" else str(pathlib.PurePosixPath(path, job_name))
 
 
 def recursive_copy(source, destination):

src/sagemaker/model_monitor/data_capture_config.py

@@ -16,8 +16,7 @@
 """
 from __future__ import print_function, absolute_import
 
-import os
-
+from sagemaker import s3
 from sagemaker.session import Session
 
 _MODEL_MONITOR_S3_PATH = "model-monitor"
@@ -67,7 +66,7 @@ def __init__(
         self.destination_s3_uri = destination_s3_uri
         if self.destination_s3_uri is None:
             sagemaker_session = sagemaker_session or Session()
-            self.destination_s3_uri = os.path.join(
+            self.destination_s3_uri = s3.s3_path_join(
                 "s3://",
                 sagemaker_session.default_bucket(),
                 _MODEL_MONITOR_S3_PATH,

src/sagemaker/model_monitor/model_monitoring.py

@@ -19,18 +19,19 @@
 import copy
 import json
 import os
+import pathlib
 import logging
 import uuid
 
 from six import string_types
 from six.moves.urllib.parse import urlparse
 from botocore.exceptions import ClientError
 
+from sagemaker import s3
 from sagemaker.exceptions import UnexpectedStatusException
 from sagemaker.model_monitor.monitoring_files import Constraints, ConstraintViolations, Statistics
 from sagemaker.network import NetworkConfig
 from sagemaker.processing import Processor, ProcessingInput, ProcessingJob, ProcessingOutput
-from sagemaker.s3 import S3Uploader
 from sagemaker.session import Session
 from sagemaker.utils import name_from_base, retries, get_ecr_image_uri_prefix
 
@@ -829,8 +830,10 @@ def _normalize_endpoint_input(self, endpoint_input):
         if isinstance(endpoint_input, string_types):
             endpoint_input = EndpointInput(
                 endpoint_name=endpoint_input,
-                destination=os.path.join(
-                    _CONTAINER_BASE_PATH, _CONTAINER_INPUT_PATH, _CONTAINER_ENDPOINT_INPUT_PATH
+                destination=str(
+                    pathlib.PurePosixPath(
+                        _CONTAINER_BASE_PATH, _CONTAINER_INPUT_PATH, _CONTAINER_ENDPOINT_INPUT_PATH
+                    )
                 ),
             )
 
@@ -862,13 +865,13 @@ def _normalize_baseline_inputs(self, baseline_inputs=None):
                 # and save the S3 uri in the ProcessingInput source.
                 parse_result = urlparse(file_input.source)
                 if parse_result.scheme != "s3":
-                    s3_uri = os.path.join(
+                    s3_uri = s3.s3_path_join(
                         "s3://",
                         self.sagemaker_session.default_bucket(),
                         self.latest_baselining_job_name,
                         file_input.input_name,
                     )
-                    S3Uploader.upload(
+                    s3.S3Uploader.upload(
                         local_path=file_input.source,
                         desired_s3_uri=s3_uri,
                         sagemaker_session=self.sagemaker_session,
@@ -889,7 +892,7 @@ def _normalize_processing_output(self, output=None):
         """
         # If the output is a string, turn it into a ProcessingOutput object.
         if isinstance(output, string_types):
-            s3_uri = os.path.join(
+            s3_uri = s3.s3_path_join(
                 "s3://",
                 self.sagemaker_session.default_bucket(),
                 self.latest_baselining_job_name,
@@ -913,7 +916,7 @@ def _normalize_monitoring_output(self, output=None):
         """
         # If the output is a string, turn it into a ProcessingOutput object.
         if output.destination is None:
-            output.destination = os.path.join(
+            output.destination = s3.s3_path_join(
                 "s3://",
                 self.sagemaker_session.default_bucket(),
                 self.monitoring_schedule_name,
@@ -934,7 +937,7 @@ def _s3_uri_from_local_path(self, path):
         """
         parse_result = urlparse(path)
         if parse_result.scheme != "s3":
-            s3_uri = os.path.join(
+            s3_uri = s3.s3_path_join(
                 "s3://",
                 self.sagemaker_session.default_bucket(),
                 _MODEL_MONITOR_S3_PATH,
@@ -943,10 +946,10 @@ def _s3_uri_from_local_path(self, path):
                 _INPUT_S3_PATH,
                 str(uuid.uuid4()),
             )
-            S3Uploader.upload(
+            s3.S3Uploader.upload(
                 local_path=path, desired_s3_uri=s3_uri, sagemaker_session=self.sagemaker_session
             )
-            path = os.path.join(s3_uri, os.path.basename(path))
+            path = s3.s3_path_join(s3_uri, os.path.basename(path))
         return path
 
     def _wait_for_schedule_changes_to_apply(self):
@@ -1094,8 +1097,10 @@ def suggest_baseline(
 
         normalized_baseline_dataset_input = self._upload_and_convert_to_processing_input(
             source=baseline_dataset,
-            destination=os.path.join(
-                _CONTAINER_BASE_PATH, _CONTAINER_INPUT_PATH, _BASELINE_DATASET_INPUT_NAME
+            destination=str(
+                pathlib.PurePosixPath(
+                    _CONTAINER_BASE_PATH, _CONTAINER_INPUT_PATH, _BASELINE_DATASET_INPUT_NAME
+                )
             ),
             name=_BASELINE_DATASET_INPUT_NAME,
         )
@@ -1105,34 +1110,44 @@ def suggest_baseline(
 
         normalized_record_preprocessor_script_input = self._upload_and_convert_to_processing_input(
             source=record_preprocessor_script,
-            destination=os.path.join(
-                _CONTAINER_BASE_PATH, _CONTAINER_INPUT_PATH, _RECORD_PREPROCESSOR_SCRIPT_INPUT_NAME
+            destination=str(
+                pathlib.PurePosixPath(
+                    _CONTAINER_BASE_PATH,
+                    _CONTAINER_INPUT_PATH,
+                    _RECORD_PREPROCESSOR_SCRIPT_INPUT_NAME,
+                )
             ),
             name=_RECORD_PREPROCESSOR_SCRIPT_INPUT_NAME,
         )
 
         record_preprocessor_script_container_path = None
         if normalized_record_preprocessor_script_input is not None:
-            record_preprocessor_script_container_path = os.path.join(
-                normalized_record_preprocessor_script_input.destination,
-                os.path.basename(record_preprocessor_script),
+            record_preprocessor_script_container_path = str(
+                pathlib.PurePosixPath(
+                    normalized_record_preprocessor_script_input.destination,
+                    os.path.basename(record_preprocessor_script),
+                )
             )
 
         normalized_post_processor_script_input = self._upload_and_convert_to_processing_input(
             source=post_analytics_processor_script,
-            destination=os.path.join(
-                _CONTAINER_BASE_PATH,
-                _CONTAINER_INPUT_PATH,
-                _POST_ANALYTICS_PROCESSOR_SCRIPT_INPUT_NAME,
+            destination=str(
+                pathlib.PurePosixPath(
+                    _CONTAINER_BASE_PATH,
+                    _CONTAINER_INPUT_PATH,
+                    _POST_ANALYTICS_PROCESSOR_SCRIPT_INPUT_NAME,
+                )
             ),
             name=_POST_ANALYTICS_PROCESSOR_SCRIPT_INPUT_NAME,
         )
 
         post_processor_script_container_path = None
         if normalized_post_processor_script_input is not None:
-            post_processor_script_container_path = os.path.join(
-                normalized_post_processor_script_input.destination,
-                os.path.basename(post_analytics_processor_script),
+            post_processor_script_container_path = str(
+                pathlib.PurePosixPath(
+                    normalized_post_processor_script_input.destination,
+                    os.path.basename(post_analytics_processor_script),
+                )
             )
 
         normalized_baseline_output = self._normalize_baseline_output(output_s3_uri=output_s3_uri)
@@ -1651,7 +1666,7 @@ def _normalize_baseline_output(self, output_s3_uri=None):
             sagemaker.processing.ProcessingOutput: The normalized ProcessingOutput object.
 
         """
-        s3_uri = output_s3_uri or os.path.join(
+        s3_uri = output_s3_uri or s3.s3_path_join(
             "s3://",
             self.sagemaker_session.default_bucket(),
             _MODEL_MONITOR_S3_PATH,
@@ -1660,7 +1675,7 @@ def _normalize_baseline_output(self, output_s3_uri=None):
             _RESULTS_S3_PATH,
         )
         return ProcessingOutput(
-            source=os.path.join(_CONTAINER_BASE_PATH, _CONTAINER_OUTPUT_PATH),
+            source=str(pathlib.PurePosixPath(_CONTAINER_BASE_PATH, _CONTAINER_OUTPUT_PATH)),
             destination=s3_uri,
             output_name=_DEFAULT_OUTPUT_NAME,
         )
@@ -1675,7 +1690,7 @@ def _normalize_monitoring_output(self, output_s3_uri=None):
             sagemaker.model_monitor.MonitoringOutput: The normalized MonitoringOutput object.
 
         """
-        s3_uri = output_s3_uri or os.path.join(
+        s3_uri = output_s3_uri or s3.s3_path_join(
             "s3://",
             self.sagemaker_session.default_bucket(),
             _MODEL_MONITOR_S3_PATH,
@@ -1684,7 +1699,8 @@ def _normalize_monitoring_output(self, output_s3_uri=None):
             _RESULTS_S3_PATH,
         )
         output = MonitoringOutput(
-            source=os.path.join(_CONTAINER_BASE_PATH, _CONTAINER_OUTPUT_PATH), destination=s3_uri
+            source=str(pathlib.PurePosixPath(_CONTAINER_BASE_PATH, _CONTAINER_OUTPUT_PATH)),
+            destination=s3_uri,
         )
 
         return output
@@ -1761,7 +1777,7 @@ def _upload_and_convert_to_processing_input(self, source, destination, name):
         parse_result = urlparse(url=source)
 
         if parse_result.scheme != "s3":
-            s3_uri = os.path.join(
+            s3_uri = s3.s3_path_join(
                 "s3://",
                 self.sagemaker_session.default_bucket(),
                 _MODEL_MONITOR_S3_PATH,
@@ -1770,7 +1786,7 @@ def _upload_and_convert_to_processing_input(self, source, destination, name):
                 _INPUT_S3_PATH,
                 name,
             )
-            S3Uploader.upload(
+            s3.S3Uploader.upload(
                 local_path=source, desired_s3_uri=s3_uri, sagemaker_session=self.sagemaker_session
             )
             source = s3_uri
@@ -1861,7 +1877,7 @@ def baseline_statistics(self, file_name=STATISTICS_JSON_DEFAULT_FILE_NAME, kms_k
         try:
             baselining_job_output_s3_path = self.outputs[0].destination
             return Statistics.from_s3_uri(
-                statistics_file_s3_uri=os.path.join(baselining_job_output_s3_path, file_name),
+                statistics_file_s3_uri=s3.s3_path_join(baselining_job_output_s3_path, file_name),
                 kms_key=kms_key,
                 sagemaker_session=self.sagemaker_session,
             )
@@ -1899,7 +1915,7 @@ def suggested_constraints(self, file_name=CONSTRAINTS_JSON_DEFAULT_FILE_NAME, km
         try:
             baselining_job_output_s3_path = self.outputs[0].destination
             return Constraints.from_s3_uri(
-                constraints_file_s3_uri=os.path.join(baselining_job_output_s3_path, file_name),
+                constraints_file_s3_uri=s3.s3_path_join(baselining_job_output_s3_path, file_name),
                 kms_key=kms_key,
                 sagemaker_session=self.sagemaker_session,
             )
@@ -2015,7 +2031,7 @@ def statistics(self, file_name=STATISTICS_JSON_DEFAULT_FILE_NAME, kms_key=None):
         try:
             baselining_job_output_s3_path = self.outputs[0].destination
             return Statistics.from_s3_uri(
-                statistics_file_s3_uri=os.path.join(baselining_job_output_s3_path, file_name),
+                statistics_file_s3_uri=s3.s3_path_join(baselining_job_output_s3_path, file_name),
                 kms_key=kms_key,
                 sagemaker_session=self.sagemaker_session,
             )
@@ -2055,7 +2071,7 @@ def constraint_violations(
         try:
             baselining_job_output_s3_path = self.outputs[0].destination
             return ConstraintViolations.from_s3_uri(
-                constraint_violations_file_s3_uri=os.path.join(
+                constraint_violations_file_s3_uri=s3.s3_path_join(
                     baselining_job_output_s3_path, file_name
                 ),
                 kms_key=kms_key,

src/sagemaker/model_monitor/monitoring_files.py

@@ -22,9 +22,8 @@
 
 from botocore.exceptions import ClientError
 
+from sagemaker import s3
 from sagemaker.session import Session
-from sagemaker.s3 import S3Downloader
-from sagemaker.s3 import S3Uploader
 
 NO_SUCH_KEY_CODE = "NoSuchKey"
 
@@ -68,7 +67,7 @@ def save(self, new_save_location_s3_uri=None):
         if new_save_location_s3_uri is not None:
             self.file_s3_uri = new_save_location_s3_uri
 
-        return S3Uploader.upload_string_as_file_body(
+        return s3.S3Uploader.upload_string_as_file_body(
             body=json.dumps(self.body_dict),
             desired_s3_uri=self.file_s3_uri,
             kms_key=self.kms_key,
@@ -119,7 +118,7 @@ def from_s3_uri(cls, statistics_file_s3_uri, kms_key=None, sagemaker_session=Non
         """
         try:
             body_dict = json.loads(
-                S3Downloader.read_file(
+                s3.S3Downloader.read_file(
                     s3_uri=statistics_file_s3_uri, sagemaker_session=sagemaker_session
                 )
             )
@@ -158,10 +157,10 @@ def from_string(
         """
         sagemaker_session = sagemaker_session or Session()
         file_name = file_name or "statistics.json"
-        desired_s3_uri = os.path.join(
+        desired_s3_uri = s3.s3_path_join(
             "s3://", sagemaker_session.default_bucket(), "monitoring", str(uuid.uuid4()), file_name
         )
-        s3_uri = S3Uploader.upload_string_as_file_body(
+        s3_uri = s3.S3Uploader.upload_string_as_file_body(
             body=statistics_file_string,
             desired_s3_uri=desired_s3_uri,
             kms_key=kms_key,
@@ -245,7 +244,7 @@ def from_s3_uri(cls, constraints_file_s3_uri, kms_key=None, sagemaker_session=No
         """
         try:
             body_dict = json.loads(
-                S3Downloader.read_file(
+                s3.S3Downloader.read_file(
                     s3_uri=constraints_file_s3_uri, sagemaker_session=sagemaker_session
                 )
             )
@@ -287,10 +286,10 @@ def from_string(
         """
         sagemaker_session = sagemaker_session or Session()
         file_name = file_name or "constraints.json"
-        desired_s3_uri = os.path.join(
+        desired_s3_uri = s3.s3_path_join(
             "s3://", sagemaker_session.default_bucket(), "monitoring", str(uuid.uuid4()), file_name
         )
-        s3_uri = S3Uploader.upload_string_as_file_body(
+        s3_uri = s3.S3Uploader.upload_string_as_file_body(
             body=constraints_file_string,
             desired_s3_uri=desired_s3_uri,
             kms_key=kms_key,
@@ -399,7 +398,7 @@ def from_s3_uri(cls, constraint_violations_file_s3_uri, kms_key=None, sagemaker_
         """
         try:
             body_dict = json.loads(
-                S3Downloader.read_file(
+                s3.S3Downloader.read_file(
                     s3_uri=constraint_violations_file_s3_uri, sagemaker_session=sagemaker_session
                 )
             )
@@ -442,10 +441,10 @@ def from_string(
         """
         sagemaker_session = sagemaker_session or Session()
         file_name = file_name or "constraint_violations.json"
-        desired_s3_uri = os.path.join(
+        desired_s3_uri = s3.s3_path_join(
             "s3://", sagemaker_session.default_bucket(), "monitoring", str(uuid.uuid4()), file_name
         )
-        s3_uri = S3Uploader.upload_string_as_file_body(
+        s3_uri = s3.S3Uploader.upload_string_as_file_body(
             body=constraint_violations_file_string,
             desired_s3_uri=desired_s3_uri,
             kms_key=kms_key,