only keep selection

Author: Andrei Neagu

services/storage/src/simcore_service_storage/simcore_s3_dsm.py

@@ -81,6 +81,7 @@
 from .modules.s3 import get_s3_client
 from .utils.s3_utils import S3TransferDataCB
 from .utils.simcore_s3_dsm_utils import (
+    UserSelection,
     compute_file_id_prefix,
     create_and_upload_export,
     create_random_export_name,
@@ -1249,7 +1250,7 @@ async def create_s3_export(
         *,
         progress_bar: ProgressBarData,
     ) -> StorageFileID:
-        source_object_keys: set[StorageFileID] = set()
+        source_object_keys: set[tuple[UserSelection, StorageFileID]] = set()
 
         # check access rights
         for object_key in object_keys:
@@ -1279,7 +1280,7 @@ async def create_s3_export(
                 self.simcore_bucket_name, object_key
             ):
                 for entry in meta_data_files:
-                    source_object_keys.add(entry.object_key)
+                    source_object_keys.add((object_key, entry.object_key))
 
         _logger.debug(
             "User selection '%s' includes '%s' files",

services/storage/src/simcore_service_storage/utils/simcore_s3_dsm_utils.py

@@ -1,10 +1,12 @@
 from contextlib import suppress
 from pathlib import Path
+from typing import TypeAlias
 from uuid import uuid4
 
 import orjson
 from aws_library.s3 import S3MetaData, SimcoreS3API
 from aws_library.s3._constants import STREAM_READER_CHUNK_SIZE
+from aws_library.s3._models import S3ObjectKey
 from models_library.api_schemas_storage.storage_schemas import S3BucketName
 from models_library.projects import ProjectID
 from models_library.projects_nodes_io import (
@@ -143,20 +145,33 @@ def create_random_export_name(user_id: UserID) -> StorageFileID:
     )
 
 
+UserSelection: TypeAlias = str
+
+
+def _strip_parent(selection: UserSelection, s3_object: S3ObjectKey) -> str:
+    selection_path = Path(selection)
+    s3_object_path = Path(s3_object)
+    if selection_path == s3_object_path:
+        return s3_object_path.name
+
+    result = s3_object_path.relative_to(selection_path)
+    return f"{result}"
+
+
 async def create_and_upload_export(
     s3_client: SimcoreS3API,
     bucket: S3BucketName,
     *,
-    source_object_keys: set[StorageFileID],
+    source_object_keys: set[tuple[UserSelection, StorageFileID]],
     destination_object_keys: StorageFileID,
     progress_bar: ProgressBarData,
 ) -> None:
     archive_entries: ArchiveEntries = [
         (
-            s3_object,
+            _strip_parent(selection, s3_object),
             await s3_client.get_bytes_streamer_from_object(bucket, s3_object),
         )
-        for s3_object in source_object_keys
+        for (selection, s3_object) in source_object_keys
     ]
 
     async with progress_bar:

services/storage/tests/unit/test_simcore_s3_dsm_utils.py

@@ -1,5 +1,12 @@
+from pathlib import Path
+
 import pytest
-from simcore_service_storage.utils.simcore_s3_dsm_utils import compute_file_id_prefix
+from aws_library.s3._models import S3ObjectKey
+from simcore_service_storage.utils.simcore_s3_dsm_utils import (
+    UserSelection,
+    _strip_parent,
+    compute_file_id_prefix,
+)
 
 
 @pytest.mark.parametrize(
@@ -19,3 +26,23 @@
 )
 def test_compute_file_id_prefix(file_id, levels, expected):
     assert compute_file_id_prefix(file_id, levels) == expected
+
+
+_FOLDERS_PATH = Path("nested/folders/path")
+
+
+@pytest.mark.parametrize(
+    "selection, s3_object, expected",
+    [
+        (Path("single_file"), Path("single_file"), "single_file"),
+        (Path("single_folder"), Path("single_folder"), "single_folder"),
+        (_FOLDERS_PATH / "folder", _FOLDERS_PATH / "folder", "folder"),
+        (_FOLDERS_PATH / "a_file.txt", _FOLDERS_PATH / "a_file.txt", "a_file.txt"),
+        (_FOLDERS_PATH, _FOLDERS_PATH / "the/actual/path", "the/actual/path"),
+    ],
+)
+def test__strip_parent(selection: Path, s3_object: Path, expected: str):
+    assert (
+        _strip_parent(UserSelection(f"{selection}"), S3ObjectKey(f"{s3_object}"))
+        == expected
+    )