Return storage alias in dataset information service response (#3)

ghga-de · Oct 11, 2024 · 5e41b4d · 5e41b4d
1 parent 5947e81
commit 5e41b4d
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 3 deletions.
diff --git a/openapi.yaml b/openapi.yaml
@@ -53,10 +53,16 @@ components:
           exclusiveMinimum: 0.0
           title: Size
           type: integer
+        storage_alias:
+          description: Alias of the storage location where the corresponding file
+            data resides in permanent storage.
+          title: Storage Alias
+          type: string
       required:
       - accession
       - size
       - sha256_hash
+      - storage_alias
       title: FileInformation
       type: object
     HTTPValidationError:

diff --git a/src/dins/core/information_service.py b/src/dins/core/information_service.py
@@ -88,6 +88,7 @@ async def register_file_information(
             accession=file.file_id,
             size=file.decrypted_size,
             sha256_hash=file.decrypted_sha256,
+            storage_alias=file.s3_endpoint_alias,
         )
         file_id = file_information.accession
 

diff --git a/src/dins/core/models.py b/src/dins/core/models.py
@@ -40,6 +40,11 @@ class FileInformation(FileAccession):
         description="SHA256 hash of the unencrypted file content encoded as hexadecimal"
         " values as produced by hashlib.hexdigest().",
     )
+    storage_alias: str = Field(
+        default=...,
+        description="Alias of the storage location where the corresponding file data resides"
+        " in permanent storage.",
+    )
 
 
 class DatasetFileAccessions(BaseModel):

diff --git a/tests/test_typical_journey.py b/tests/test_typical_journey.py
@@ -80,14 +80,17 @@
 )
 
 INCOMING_FILE_PAYLOAD_2 = INCOMING_FILE_PAYLOAD.model_copy(
-    update={"file_id": FILE_ID_2}
+    update={"file_id": FILE_ID_2, "s3_endpoint_alias": "test-node-2"}
 )
 INCOMING_FILE_PAYLOAD_3 = INCOMING_FILE_PAYLOAD.model_copy(
-    update={"file_id": FILE_ID_3}
+    update={"file_id": FILE_ID_3, "s3_endpoint_alias": "test-node-3"}
 )
 
 FILE_INFORMATION = models.FileInformation(
-    accession=FILE_ID_1, sha256_hash=DECRYPTED_SHA256, size=DECRYPTED_SIZE
+    accession=FILE_ID_1,
+    sha256_hash=DECRYPTED_SHA256,
+    size=DECRYPTED_SIZE,
+    storage_alias="test-node",
 )
 
 pytestmark = pytest.mark.asyncio()
@@ -227,11 +230,13 @@ async def test_dataset_information_journey(
             "accession": FILE_ID_1,
             "size": DECRYPTED_SIZE,
             "sha256_hash": DECRYPTED_SHA256,
+            "storage_alias": "test-node",
         },
         {
             "accession": FILE_ID_2,
             "size": DECRYPTED_SIZE,
             "sha256_hash": DECRYPTED_SHA256,
+            "storage_alias": "test-node-2",
         },
     ]
 
@@ -256,16 +261,19 @@ async def test_dataset_information_journey(
             "accession": FILE_ID_1,
             "size": DECRYPTED_SIZE,
             "sha256_hash": DECRYPTED_SHA256,
+            "storage_alias": "test-node",
         },
         {
             "accession": FILE_ID_2,
             "size": DECRYPTED_SIZE,
             "sha256_hash": DECRYPTED_SHA256,
+            "storage_alias": "test-node-2",
         },
         {
             "accession": FILE_ID_3,
             "size": DECRYPTED_SIZE,
             "sha256_hash": DECRYPTED_SHA256,
+            "storage_alias": "test-node-3",
         },
     ]