fix leading slash case

Author: laurenyu

src/sagemaker/local/utils.py

@@ -14,12 +14,13 @@
 from __future__ import absolute_import
 
 import os
-import pathlib
 import shutil
 
 from distutils.dir_util import copy_tree
 from six.moves.urllib.parse import urlparse
 
+from sagemaker import s3
+
 
 def copy_directory_structure(destination_directory, relative_path):
     """Create all the intermediate directories required for relative_path to
@@ -63,8 +64,8 @@ def move_to_destination(source, destination, job_name, sagemaker_session):
         final_uri = destination
     elif parsed_uri.scheme == "s3":
         bucket = parsed_uri.netloc
-        path = _create_s3_prefix(parsed_uri.path, job_name)
-        final_uri = "s3://%s/%s" % (bucket, path)
+        path = s3.s3_path_join(parsed_uri.path, job_name)
+        final_uri = s3.s3_path_join(bucket, path)
         sagemaker_session.upload_data(source, bucket, path)
     else:
         raise ValueError("Invalid destination URI, must be s3:// or file://, got: %s" % destination)
@@ -73,21 +74,6 @@ def move_to_destination(source, destination, job_name, sagemaker_session):
     return final_uri
 
 
-def _create_s3_prefix(path, job_name):
-    """Constructs a path out of the given path and job name to be
-    used as an S3 prefix.
-
-    Args:
-        path (str): the original path. If the path is only ``"/"``,
-            then it is ignored.
-        job_name (str): the job name to be appended to the path.
-
-    Returns:
-        str: an S3 prefix of the form ``"path/job_name"``
-    """
-    return job_name if path == "" else str(pathlib.PurePosixPath(path, job_name))
-
-
 def recursive_copy(source, destination):
     """A wrapper around distutils.dir_util.copy_tree but won't throw any
     exception when the source directory does not exist.

src/sagemaker/s3.py

@@ -52,11 +52,11 @@ def s3_path_join(*args):
     Returns:
         str: The joined string.
     """
-    path = str(pathlib.PurePosixPath(*args))
-    if path.startswith("s3:/"):
-        return path.replace("s3:/", "s3://")
+    if args[0].startswith("s3://"):
+        path = str(pathlib.PurePosixPath(*args[1:])).lstrip("/")
+        return str(pathlib.PurePosixPath(args[0], path)).replace("s3:/", "s3://")
 
-    return path
+    return str(pathlib.PurePosixPath(*args)).lstrip("/")
 
 
 class S3Uploader(object):

tests/unit/test_s3.py

@@ -112,7 +112,9 @@ def test_path_join():
     test_cases = (
         ("foo/bar", ("foo", "bar")),
         ("foo/bar", ("foo/", "bar")),
+        ("foo/bar", ("/foo/", "bar")),
         ("s3://foo/bar", ("s3://", "foo", "bar")),
+        ("s3://foo/bar", ("s3://", "/foo", "bar")),
         ("s3://foo/bar", ("s3://foo", "bar")),
     )