From 136300ed0d40a7f35ff3abc4de4c688cf598368a Mon Sep 17 00:00:00 2001
From: Sven Marnach <sven@mozilla.com>
Date: Thu, 11 Jul 2024 14:39:37 +0200
Subject: [PATCH] bug-1905455: Enable anonymous access to GCS buckets.

---
 tecken/ext/gcs/storage.py | 33 +++++++++++++++++++--------------
 1 file changed, 19 insertions(+), 14 deletions(-)

diff --git a/tecken/ext/gcs/storage.py b/tecken/ext/gcs/storage.py
index 00264c355..dc8c352c5 100644
--- a/tecken/ext/gcs/storage.py
+++ b/tecken/ext/gcs/storage.py
@@ -8,11 +8,12 @@
 from urllib.parse import quote, urlparse
 
 from django.conf import settings
-
 from google.api_core.client_options import ClientOptions
-from google.api_core.exceptions import ClientError, NotFound
+from google.api_core.exceptions import ClientError
 from google.cloud import storage
+from requests.exceptions import RequestException
 
+from tecken.librequests import session_with_retries
 from tecken.libstorage import ObjectMetadata, StorageBackend, StorageError
 
 
@@ -52,13 +53,7 @@ def _get_client(self) -> storage.Client:
 
     def _get_bucket(self) -> storage.Bucket:
         """Return a thread-local low-level storage bucket client."""
-        if not hasattr(self.clients, "bucket"):
-            client = self._get_client()
-            try:
-                self.clients.bucket = client.get_bucket(self.name, timeout=self.timeout)
-            except NotFound as exc:
-                raise StorageError(self) from exc
-        return self.clients.bucket
+        return self._get_client().bucket(self.name)
 
     def exists(self) -> bool:
         """Check that this storage exists.
@@ -67,13 +62,23 @@ def exists(self) -> bool:
 
         :raises StorageError: an unexpected backend-specific error was raised
         """
+        session = session_with_retries()
+        if self.endpoint_url.startswith("http://gcs-emulator"):
+            # NOTE(smarnach): The GCS emulator does not support HEAD requests. Moreover, the
+            # simpler public endpoint used below will throw 500s if the bucket doesn't exists,
+            # tripping up the retry behaviour. We can't use this endpoint for Google's API,
+            # since it requires authentication, and we want to be able to detect the existence
+            # of public buckets anonymously (bug 1905455).
+            method = "GET"
+            url = f"{self.endpoint_url}/storage/v1/b/{self.name}"
+        else:
+            method = "HEAD"
+            url = f"{self.endpoint_url}/{self.name}"
         try:
-            self._get_bucket()
-        except StorageError:
-            return False
-        except ClientError as exc:
+            response = session.request(method, url)
+        except RequestException as exc:
             raise StorageError(self) from exc
-        return True
+        return response.status_code == 200
 
     def get_object_metadata(self, key: str) -> Optional[ObjectMetadata]:
         """Return object metadata for the object with the given key.