Merge tag 'debian/1.4.4-1_swh1' into debian/buster-swh

PKG-INFO

 Metadata-Version: 2.1
 Name: swh.deposit
-Version: 1.4.3
+Version: 1.4.4
 Summary: Software Heritage Deposit Server
 Home-page: https://forge.softwareheritage.org/source/swh-deposit/
 Author: Software Heritage developers

debian/changelog

-swh-deposit (1.4.3-1~swh1~bpo10+1) buster-swh; urgency=medium
+swh-deposit (1.4.4-1~swh1) unstable-swh; urgency=medium
 
-  * Rebuild for buster-swh
+  * New upstream release 1.4.4     - (tagged by Antoine R. Dumont
+    (@ardumont) <ardumont@softwareheritage.org> on 2023-11-24 12:40:08
+    +0100)
+  * Upstream changes:     - v1.4.4     - azure blobstorage: Reset blob's
+    content-encoding to allow check
 
- -- Software Heritage autobuilder (on jenkins-debian1) <jenkins@jenkins-debian1.internal.softwareheritage.org>  Fri, 24 Nov 2023 09:03:34 +0000
+ -- Software Heritage autobuilder (on jenkins-debian1) <jenkins@jenkins-debian1.internal.softwareheritage.org>  Fri, 24 Nov 2023 11:46:38 +0000
 
 swh-deposit (1.4.3-1~swh1) unstable-swh; urgency=medium
 

mypy.ini

@@ -44,3 +44,9 @@ ignore_missing_imports = True
 
 [mypy-swh.storage.*]
 ignore_missing_imports = True
+
+[mypy-storages.*]
+ignore_missing_imports = True
+
+[mypy-azure.*]
+ignore_missing_imports = True

swh.deposit.egg-info/PKG-INFO

 Metadata-Version: 2.1
 Name: swh.deposit
-Version: 1.4.3
+Version: 1.4.4
 Summary: Software Heritage Deposit Server
 Home-page: https://forge.softwareheritage.org/source/swh-deposit/
 Author: Software Heritage developers

swh/deposit/api/private/deposit_check.py

@@ -125,6 +125,11 @@ class APIChecks(APIPrivateView, APIGet, DepositReadMixin):
             # Use python's File api which is consistent across different types of
             # storage backends (e.g. file, azure, ...)
 
+            # I did not find any other) workaround for azure blobstorage use, noop
+            # otherwise
+            reset_content_settings_if_needed(archive)
+            # FIXME: ^ Implement a better way (after digging into django-storages[azure]
+
             with archive.open("rb") as archive_fp:
                 try:
                     with zipfile.ZipFile(archive_fp) as zip_fp:
@@ -214,3 +219,35 @@ class APIChecks(APIPrivateView, APIGet, DepositReadMixin):
         deposit.save()
 
         return status.HTTP_200_OK, response, "application/json"
+
+
+def reset_content_settings_if_needed(archive) -> None:
+    """This resets the content_settings on the associated blob stored in an azure
+    blobstorage. This prevents the correct reading of the file and failing the checks
+    for no good reason.
+
+    """
+    try:
+        from storages.backends.azure_storage import AzureStorage
+    except ImportError:
+        return None
+
+    if not isinstance(archive.storage, AzureStorage):
+        return None
+
+    from azure.storage.blob import ContentSettings
+
+    blob_client = archive.storage.client.get_blob_client(archive.name)
+
+    # Get the existing blob properties
+    properties = blob_client.get_blob_properties()
+
+    # reset content encoding in the settings
+    content_settings = dict(properties.content_settings)
+    content_settings["content_encoding"] = ""
+
+    # Set the content_type and content_language headers, and populate the remaining
+    # headers from the existing properties
+    blob_headers = ContentSettings(**content_settings)
+
+    blob_client.set_http_headers(blob_headers)