Make oras-py behave the same way as oras-go for deciding whether to unpack a layer

CHANGELOG.md

@@ -17,6 +17,7 @@ The versions coincide with releases on pip. Only major versions will be released
  - check for blob existence before uploading (0.2.26)
  - fix get_tags for ECR when limit is None, closes issue [173](https://github.com/oras-project/oras-py/issues/173)
  - fix empty token for anon tokens to work, closes issue [167](https://github.com/oras-project/oras-py/issues/167)
+ - use same annotation as oras-go for determining whether to unpack a layer or not [119](https://github.com/oras-project/oras-py/issues/119)
  - retry on 500 (0.2.25)
  - align provider config_path type annotations (0.2.24)
  - add missing prefix property to auth backend (0.2.23)

oras/provider.py

@@ -781,23 +781,24 @@ def push(
  f"Blob {blob} is not in the present working directory context."
  )
 
- # Save directory or blob name before compressing
- blob_name = os.path.basename(blob)
-
  # If it's a directory, we need to compress
- cleanup_blob = False
- if os.path.isdir(blob):
- blob = oras.utils.make_targz(blob)
- cleanup_blob = True
+ is_dir_layer = os.path.isdir(blob)
 
+ blob_to_use_for_layer = (
+ oras.utils.make_targz(blob) if is_dir_layer else blob
+ )
  # Create a new layer from the blob
- layer = oras.oci.NewLayer(blob, is_dir=cleanup_blob, media_type=media_type)
+ layer = oras.oci.NewLayer(
+ blob_to_use_for_layer, is_dir=is_dir_layer, media_type=media_type
+ )
  annotations = annotset.get_annotations(blob)
 
  # Always strip blob_name of path separator
- layer["annotations"] = {
- oras.defaults.annotation_title: blob_name.strip(os.sep)
- }
+ layer["annotations"] = {oras.defaults.annotation_title: blob.strip(os.sep)}
+
+ if is_dir_layer:
+ layer["annotations"][oras.defaults.annotation_unpack] = "True"
+
  if annotations:
  layer["annotations"].update(annotations)
 
@@ -807,7 +808,7 @@ def push(
 
  # Upload the blob layer
  response = self.upload_blob(
- blob,
+ blob_to_use_for_layer,
  container,
  layer,
  do_chunked=do_chunked,
@@ -816,8 +817,8 @@ def push(
  self._check_200_response(response)
 
  # Do we need to cleanup a temporary targz?
- if cleanup_blob and os.path.exists(blob):
- os.remove(blob)
+ if is_dir_layer and os.path.exists(blob_to_use_for_layer):
+ os.remove(blob_to_use_for_layer)
 
  # Add annotations to the manifest, if provided
  manifest_annots = annotset.get_annotations("$manifest") or {}
@@ -872,22 +873,21 @@ def pull(
  allowed_media_type: Optional[List] = None,
  overwrite: bool = True,
  outdir: Optional[str] = None,
+ skip_unpack: bool = False,
  ) -> List[str]:
  """
  Pull an artifact from a target
 
+ :param target: target location to pull from
+ :type target: str
  :param config_path: path to a config file
  :type config_path: str
  :param allowed_media_type: list of allowed media types
  :type allowed_media_type: list or None
  :param overwrite: if output file exists, overwrite
  :type overwrite: bool
- :param manifest_config_ref: save manifest config to this file
- :type manifest_config_ref: str
  :param outdir: output directory path
  :type outdir: str
- :param target: target location to pull from
- :type target: str
  """
  container = self.get_container(target)
  self.auth.load_configs(
@@ -899,9 +899,10 @@ def pull(
 
  files = []
  for layer in manifest.get("layers", []):
- filename = (layer.get("annotations") or {}).get(
- oras.defaults.annotation_title
- )
+ annotations: dict = layer.get("annotations", {})
+ filename = annotations.get(oras.defaults.annotation_title)
+ # A directory will need to be uncompressed and moved
+ unpack_layer = annotations.get(oras.defaults.annotation_unpack, False)
 
  # If we don't have a filename, default to digest. Hopefully does not happen
  if not filename:
@@ -916,8 +917,7 @@ def pull(
  )
  continue
 
- # A directory will need to be uncompressed and moved
- if layer["mediaType"] == oras.defaults.default_blob_dir_media_type:
+ if unpack_layer:
  targz = oras.utils.get_tmpfile(suffix=".tar.gz")
  self.download_blob(container, layer["digest"], targz)
 
@@ -927,7 +927,7 @@ def pull(
  # Anything else just extracted directly
  else:
  self.download_blob(container, layer["digest"], outfile)
- logger.info(f"Successfully pulled {outfile}.")
+ logger.info(f"Successfully pulled {outfile}")
  files.append(outfile)
  return files
 

oras/tests/test_oras.py

@@ -2,12 +2,14 @@
 __copyright__ = "Copyright The ORAS Authors."
 __license__ = "Apache-2.0"
 
+import json
 import os
 import shutil
 
 import pytest
 
 import oras.client
+import oras.oci
 
 here = os.path.abspath(os.path.dirname(__file__))
 
@@ -161,6 +163,29 @@ def test_directory_push_pull(tmp_path, registry, credentials, target_dir):
  assert "artifact.txt" in os.listdir(files[0])
 
 
+@pytest.mark.with_auth(False)
+def test_directory_push_pull_skip_unpack(tmp_path, registry, credentials, target_dir):
+ """
+ Test push and pull for directory
+ """
+ client = oras.client.OrasClient(hostname=registry, insecure=True)
+
+ # Test upload of a directory with an annotation file setting oras.defaults.annotation_unpack to False
+ upload_dir = os.path.join(here, "upload_data")
+ annotation_file = os.path.join(here, "annotations.json")
+ with open(annotation_file, "w") as f:
+ json.dump({str(upload_dir):{"oras.defaults.annotation_unpack": "False"}}, f)
+
+ res = client.push(files=[upload_dir], target=target_dir, annotation_file=annotation_file)
+ assert res.status_code == 201
+ files = client.pull(target=target_dir, outdir=tmp_path)
+
+ assert len(files) == 1
+ assert os.path.basename(files[0]) == os.path.basename(upload_dir)
+ assert str(tmp_path) in files[0]
+ assert os.path.exists(files[0])
+
+
 @pytest.mark.with_auth(True)
 def test_directory_push_pull_selfsigned_auth(
  tmp_path, registry, credentials, target_dir