make repository non-required for ArchiveService, add bulk delete to ArchiveService, remove default value from paths

Author: nora-shap

shared/api_archive/archive.py

@@ -64,7 +64,10 @@ def __init__(self, repository, ttl=None):
         self.ttl = ttl or int(get_config("services", "minio", "ttl", default=self.ttl))
 
         self.storage = StorageService()
-        self.storage_hash = self.get_archive_hash(repository)
+        if repository:
+            self.storage_hash = self.get_archive_hash(repository)
+        else:
+            self.storage_hash = None
 
     @classmethod
     def get_archive_hash(cls, repository) -> str:
@@ -98,6 +101,8 @@ def write_json_data_to_storage(
         *,
         encoder=ReportEncoder,
     ):
+        if not self.storage_hash:
+            raise ValueError("No hash key provided")
         if commit_id is None:
             # Some classes don't have a commit associated with them
             # For example Pull belongs to multiple commits.
@@ -147,20 +152,29 @@ def read_file(self, path: str) -> str:
         return contents.decode()
 
     @sentry_sdk.trace
-    def delete_file(self, path):
+    def delete_file(self, path: str) -> None:
         """
         Generic method to delete a file from the archive.
         """
         self.storage.delete_file(self.root, path)
 
+    @sentry_sdk.trace
+    def delete_files(self, paths: list[str]) -> None:
+        """
+        Generic method to delete files from the archive.
+        """
+        self.storage.delete_files(bucket_name=self.root, paths=paths)
+
     def read_chunks(self, commit_sha: str) -> str:
         """
         Convenience method to read a chunks file from the archive.
         """
+        if not self.storage_hash:
+            raise ValueError("No hash key provided")
         path = MinioEndpoints.chunks.get_path(
             version="v4", repo_hash=self.storage_hash, commitid=commit_sha
         )
         return self.read_file(path)
 
-    def create_presigned_put(self, path):
+    def create_presigned_put(self, path: str) -> str:
         return self.storage.create_presigned_put(self.root, path, self.ttl)

shared/django_apps/utils/model_utils.py

@@ -115,7 +115,7 @@ def __get__(self, obj, objtype=None):
         if cached_value:
             return cached_value
         db_field = getattr(obj, self.db_field_name)
-        if db_field is not None:
+        if not db_field:
             value = self.rehydrate_fn(obj, db_field)
         else:
             value = self._get_value_from_archive(obj)

shared/storage/aws.py

@@ -147,7 +147,7 @@ def delete_file(self, bucket_name, path):
         except ClientError:
             raise
 
-    def delete_files(self, bucket_name, paths=[]):
+    def delete_files(self, bucket_name: str, paths: list[str]) -> list[bool]:
         """Batch deletes a list of files from a given bucket
 
         Note:

shared/storage/fallback.py

@@ -89,7 +89,7 @@ def delete_file(self, bucket_name, path):
         second_deletion = self.fallback_service.delete_file(bucket_name, path)
         return first_deletion and second_deletion
 
-    def delete_files(self, bucket_name, paths=[]):
+    def delete_files(self, bucket_name: str, paths: list[str]) -> list[bool]:
         """Batch deletes a list of files from a given bucket
             (what happens to the files that don't exist?)
 

shared/storage/memory.py

@@ -123,7 +123,7 @@ def delete_file(self, bucket_name, path):
             raise FileNotInStorageError()
         return True
 
-    def delete_files(self, bucket_name, paths=[]):
+    def delete_files(self, bucket_name: str, paths: list[str]) -> list[bool]:
         """Batch deletes a list of files from a given bucket
             (what happens to the files that don't exist?)
 

shared/storage/minio.py

@@ -233,21 +233,21 @@ def read_file(
         deletion, returns a ResponseError otherwise.
     """
 
-    def delete_file(self, bucket_name, url):
+    def delete_file(self, bucket_name: str, path: str) -> bool:
         try:
-            # delete a file given a bucket name and a url
-            self.minio_client.remove_object(bucket_name, url)
+            # delete a file given a bucket name and a path
+            self.minio_client.remove_object(bucket_name, path)
             return True
         except MinioException:
             raise
 
-    def delete_files(self, bucket_name, urls=[]):
+    def delete_files(self, bucket_name: str, paths: list[str]) -> list[bool]:
         try:
             for del_err in self.minio_client.remove_objects(
-                bucket_name, [DeleteObject(url) for url in urls]
+                bucket_name, [DeleteObject(path) for path in paths]
             ):
                 print("Deletion error: {}".format(del_err))
-            return [True] * len(urls)
+            return [True] * len(paths)
         except MinioException:
             raise