upload as image

Signed-off-by: Isabella do Amaral <[email protected]>

Makefile

@@ -25,7 +25,7 @@ test:
 
 .PHONY: test-e2e
 test-e2e:
-	poetry run pytest --e2e -s -x -rA
+	poetry run pytest --e2e -s -x -rA -v
 
 .PHONY: test-e2e-model-registry
 test-e2e-model-registry:

omlmd/cli.py

@@ -73,6 +73,11 @@ def crawl(plain_http: bool, targets: tuple[str]):
     required=True,
     type=click.Path(path_type=Path, exists=True, resolve_path=True),
 )
+@click.option(
+    "--as-artifact",
+    is_flag=True,
+    help="Push as an artifact (default is as a blob)",
+)
 @cloup.option_group(
     "Metadata options",
     cloup.option(
@@ -88,6 +93,7 @@ def push(
     plain_http: bool,
     target: str,
     path: Path,
+    as_artifact: bool,
     metadata: Path | None,
     empty_metadata: bool,
 ):
@@ -96,4 +102,6 @@ def push(
     if empty_metadata:
         logger.warning(f"Pushing to {target} with empty metadata.")
     md = deserialize_mdfile(metadata) if metadata else {}
-    click.echo(Helper.from_default_registry(plain_http).push(target, path, **md))
+    click.echo(
+        Helper.from_default_registry(plain_http).push(target, path, as_artifact, **md)
+    )

omlmd/constants.py

@@ -1,3 +1,6 @@
+from oras.defaults import default_blob_media_type
+
 FILENAME_METADATA_JSON = "model_metadata.omlmd.json"
-MIME_APPLICATION_CONFIG = "application/x-config"
 MIME_APPLICATION_MLMODEL = "application/x-mlmodel"
+MIME_BLOB = default_blob_media_type
+MIME_MANIFEST_CONFIG = "application/vnd.oci.image.config.v1+json"

omlmd/helpers.py

@@ -2,6 +2,8 @@
 
 import logging
 import os
+import platform
+import tarfile
 import urllib.request
 from collections.abc import Sequence
 from dataclasses import dataclass, field
@@ -10,8 +12,9 @@
 
 from .constants import (
     FILENAME_METADATA_JSON,
-    MIME_APPLICATION_CONFIG,
     MIME_APPLICATION_MLMODEL,
+    MIME_BLOB,
+    MIME_MANIFEST_CONFIG,
 )
 from .listener import Event, Listener, PushEvent
 from .model_metadata import ModelMetadata
@@ -41,12 +44,16 @@ def push(
         self,
         target: str,
         path: Path | str,
+        as_artifact: bool = False,
         **kwargs,
     ):
         owns_meta = True
         if isinstance(path, str):
             path = Path(path)
 
+        manifest_cfg = {"architecture": platform.machine(), "os": "linux"}
+
+        kwargs.update(manifest_cfg)
         meta_path = path.parent / FILENAME_METADATA_JSON
         if not kwargs and meta_path.exists():
             owns_meta = False
@@ -67,11 +74,24 @@ def push(
             model_metadata = ModelMetadata.from_dict(kwargs)
             meta_path.write_text(model_metadata.to_json())
 
-        config = f"{meta_path}:{MIME_APPLICATION_CONFIG}"
-        files = [
-            f"{path}:{MIME_APPLICATION_MLMODEL}",
-            config,
-        ]
+        config = f"{meta_path}:{MIME_MANIFEST_CONFIG}"
+
+        owns_tar = False
+        tar = None
+        if not as_artifact:
+            tar = path.parent / f"{path.stem}.tar"
+            if not tar.exists():
+                owns_tar = True
+                with tarfile.open(tar, "w") as tf:
+                    tf.add(path, arcname=path.name)
+            files = [
+                f"{tar}:{MIME_BLOB}",
+            ]
+        else:
+            files = [
+                f"{path}:{MIME_APPLICATION_MLMODEL}",
+            ]
+
         try:
             # print(target, files, model_metadata.to_annotations_dict())
             result = self._registry.push(
@@ -88,6 +108,9 @@ def push(
         finally:
             if owns_meta:
                 meta_path.unlink()
+            if owns_tar:
+                assert isinstance(tar, Path)
+                tar.unlink()
 
     def pull(
         self, target: str, outdir: Path | str, media_types: Sequence[str] | None = None

omlmd/listener.py

@@ -10,9 +10,7 @@
 
 
 class Listener(ABC):
-    """
-    TODO: not yet settled for multi-method or current single update method.
-    """
+    # TODO: not yet settled for multi-method or current single update method.
 
     @abstractmethod
     def update(self, source: t.Any, event: Event) -> None:

tests/test_e2e_model_registry.py

@@ -77,7 +77,7 @@ def update(self, source: Helper, event: Event) -> None:
     assert mv
     assert mv.description == "Lorem ipsum"
     assert mv.author == "John Doe"
-    assert mv.custom_properties == {"accuracy": 0.987}
+    assert mv.custom_properties == {"accuracy": accuracy_value}
 
     ma = model_registry.get_model_artifact("mnist", v)
     assert ma

tests/test_helpers.py

@@ -1,19 +1,27 @@
+import io
 import json
 import subprocess
+import tarfile
 import tempfile
 import typing as t
 from hashlib import sha256
 from pathlib import Path
 
 import pytest
 
-from omlmd.constants import MIME_APPLICATION_MLMODEL
+from omlmd.constants import MIME_BLOB
 from omlmd.helpers import Helper
 from omlmd.listener import Event, Listener
 from omlmd.model_metadata import ModelMetadata, deserialize_mdfile
 from omlmd.provider import OMLMDRegistry
 
 
+def untar(tar: Path, out: Path):
+    out.write_bytes(
+        t.cast(io.BufferedReader, tarfile.open(tar, "r").extractfile(tar.stem)).read()
+    )
+
+
 def test_call_push_using_md_from_file(mocker):
     helper = Helper()
     mocker.patch.object(helper, "push", return_value=None)
@@ -100,12 +108,33 @@ def test_push_pull_chunked(tmp_path, target):
 
         omlmd.push(target, temp, **md)
         omlmd.pull(target, tmp_path)
-        assert len(list(tmp_path.iterdir())) == 3
-        assert tmp_path.joinpath(temp.name).stat().st_size == base_size
+        files = list(tmp_path.iterdir())
+        print(files)
+        assert len(files) == 1
+        print(tmp_path)
+        out = tmp_path.joinpath(temp.name)
+        untar(out.with_suffix(".tar"), out)
+        assert temp.stat().st_size == base_size
     finally:
         temp.unlink()
 
 
+@pytest.mark.e2e
+def test_e2e_push_pull_as_artifact(tmp_path, target):
+    omlmd = Helper()
+    omlmd.push(
+        target,
+        Path(__file__).parent / ".." / "README.md",
+        as_artifact=True,
+        name="mnist",
+        description="Lorem ipsum",
+        author="John Doe",
+        accuracy=0.987,
+    )
+    omlmd.pull(target, tmp_path)
+    assert len(list(tmp_path.iterdir())) == 1
+
+
 @pytest.mark.e2e
 def test_e2e_push_pull(tmp_path, target):
     omlmd = Helper()
@@ -118,7 +147,7 @@ def test_e2e_push_pull(tmp_path, target):
         accuracy=0.987,
     )
     omlmd.pull(target, tmp_path)
-    assert len(list(tmp_path.iterdir())) == 3
+    assert len(list(tmp_path.iterdir())) == 1
 
 
 @pytest.mark.e2e
@@ -132,7 +161,7 @@ def test_e2e_push_pull_with_filters(tmp_path, target):
         author="John Doe",
         accuracy=0.987,
     )
-    omlmd.pull(target, tmp_path, media_types=[MIME_APPLICATION_MLMODEL])
+    omlmd.pull(target, tmp_path, media_types=[MIME_BLOB])
     assert len(list(tmp_path.iterdir())) == 1
 
 
@@ -155,10 +184,11 @@ def test_e2e_push_pull_column(tmp_path, target):
 
         omlmd.push(target, temp, **md)
         omlmd.pull(target, tmp_path)
-        with open(tmp_path.joinpath(temp.name), "r") as f:
-            pulled = f.read()
-            assert pulled == content
-            pulled_sha = sha256(pulled.encode("utf-8")).hexdigest()
-            assert pulled_sha == content_sha
+        out = tmp_path.joinpath(temp.name)
+        untar(out.with_suffix(".tar"), out)
+        pulled = out.read_text()
+        assert pulled == content
+        pulled_sha = sha256(pulled.encode("utf-8")).hexdigest()
+        assert pulled_sha == content_sha
     finally:
         temp.unlink()