From 3bb70e8a7757574096b01f2e962bd702c92b1a75 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Paul=20M=C3=BCller?= <dev@craban.de>
Date: Thu, 7 Mar 2024 09:47:19 +0100
Subject: [PATCH] feat: add convenience method `get_s3_attributes_for_artifact`

---
 CHANGELOG           |  2 ++
 dcor_shared/s3cc.py | 75 +++++++++++++++++++++++++++++++++++++++++++++
 tests/test_s3cc.py  | 19 ++++++++++++
 3 files changed, 96 insertions(+)

diff --git a/CHANGELOG b/CHANGELOG
index 0cfc9c0..8c01d52 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,3 +1,5 @@
+0.9.1
+ - feat: add convenience method `get_s3_attributes_for_artifact`
 0.9.0
  - feat: introduce `testing.make_dataset_via_s3` and
    `testing.make_resource_via_s3` which uploads resources via S3
diff --git a/dcor_shared/s3cc.py b/dcor_shared/s3cc.py
index 455960b..baa6263 100644
--- a/dcor_shared/s3cc.py
+++ b/dcor_shared/s3cc.py
@@ -61,6 +61,81 @@ def create_presigned_url(
                                    filename=filename)
 
 
+def get_s3_attributes_for_artifact(
+        resource_id: str,
+        artifact: Literal["condensed", "preview", "resource"] = "resource"):
+    """Return all attribute for an artifact in the S3 object store
+
+    Returns
+    -------
+    meta: dict
+        Metadata dictionary with the keys "etag", "server", "size",
+        and "success".
+    """
+    bucket_name, object_name = get_s3_bucket_object_for_artifact(
+        resource_id=resource_id, artifact=artifact)
+    s3_client, _, _ = s3.get_s3()
+    attr_info = s3_client.head_object(Bucket=bucket_name, Key=object_name)
+    # Example output from MinIO::
+    #
+    #     {'AcceptRanges': 'bytes',
+    #      'ContentLength': 904729,
+    #      'ContentType': 'application/octet-stream',
+    #      'ETag': '"108d47e80f3e5f35110493b1fdcd30d5"',
+    #      'LastModified': datetime.datetime(2024, 3, 7, 8, 15,
+    #                                        tzinfo=tzutc()),
+    #      'Metadata': {},
+    #      'ResponseMetadata': {
+    #         'HTTPHeaders': {
+    #             'accept-ranges': 'bytes',
+    #             'content-length': '904729',
+    #             'content-type': 'application/octet-stream',
+    #             'date': 'Thu, 07 Mar 2024 08:15:02 GMT',
+    #             'etag': '"108d47e80f3e5f35110493b1fdcd30d5"',
+    #             'last-modified': 'Thu, 07 Mar 2024 '
+    #                              '08:15:00 GMT',
+    #             'server': 'MinIO',
+    #             'strict-transport-security': 'max-age=31536000; '
+    #                                          'includeSubDomains',
+    #             'vary': 'Origin, Accept-Encoding',
+    #             'x-amz-id-2': 'dd9025bab4ad464b049177c95eb6e...',
+    #             'x-amz-request-id': '17BA6D680CB67A2C',
+    #             'x-amz-tagging-count': '1',
+    #             'x-content-type-options': 'nosniff',
+    #             'x-xss-protection': '1; mode=block'},
+    #         'HTTPStatusCode': 200,
+    #         'HostId': 'dd9025bab4ad464b049177c95eb6ebf3...',
+    #         'RequestId': '17BA6D680CB67A2C',
+    #         'RetryAttempts': 0}
+    #      }
+    meta = {}
+    for key, funcs in [
+        ("etag", [lambda m: m.get("ETag"),
+                  lambda m: m.get("ResponseMetadata",
+                                  {}).get("HTTPHeaders",
+                                          {}).get("etag"),
+                  ]),
+        ("server", [lambda m: m.get("ResponseMetadata",
+                                    {}).get("HTTPHeaders",
+                                            {}).get("server", "unknown")
+                    ]),
+        ("size", [lambda m: m.get("ContentLength"),
+                  lambda m: m.get("ResponseMetadata",
+                                  {}).get("HTTPHeaders",
+                                          {}).get("content-length"),
+                  ]),
+        ("success", [lambda m: m.get("ResponseMetadata",
+                                     {}).get("HTTPStatusCode", 404) == 200
+                     ]),
+    ]:
+        for fn in funcs:
+            val = fn(attr_info)
+            if val is not None:
+                meta[key] = val
+                break
+    return meta
+
+
 def get_s3_bucket_object_for_artifact(
         resource_id: str,
         artifact: Literal["condensed", "preview", "resource"] = "resource"):
diff --git a/tests/test_s3cc.py b/tests/test_s3cc.py
index 9cd5189..5ff1e76 100644
--- a/tests/test_s3cc.py
+++ b/tests/test_s3cc.py
@@ -115,6 +115,25 @@ def test_create_presigned_url(enqueue_job_mock, tmp_path):
         "490efdf5d9bb4cd4b2a6bcf2fe54d4dc201c38530140bcb168980bf8bf846c73"
 
 
+@pytest.mark.ckan_config('ckan.plugins', 'dcor_schemas')
+@pytest.mark.usefixtures('clean_db', 'with_request_context')
+@mock.patch('ckan.plugins.toolkit.enqueue_job',
+            side_effect=synchronous_enqueue_job)
+def test_get_s3_attributes_for_artifact(enqueue_job_mock):
+    rid, _, _, org_dict = setup_s3_resource_on_ckan()
+
+    # Make sure the resource exists
+    res_dict = helpers.call_action("resource_show", id=rid)
+    assert res_dict["id"] == rid, "sanity check"
+
+    # get the size
+    meta = s3cc.get_s3_attributes_for_artifact(rid)
+    assert meta["size"] == 904729
+    assert meta["success"]
+    assert meta["etag"]
+    assert meta["server"]
+
+
 @pytest.mark.ckan_config('ckan.plugins', 'dcor_schemas')
 @pytest.mark.usefixtures('clean_db', 'with_request_context')
 @mock.patch('ckan.plugins.toolkit.enqueue_job',