Fix ACL errors for newly created and pre-existing blobs

[RachitSharma2001]

This fixes #944 and fixes #945.

The following two python snippets no longer crash and instead give the correct output. The first one is for a pre-existing blob (thus fixing #944):

os.environ["STORAGE_EMULATOR_HOST"] = "http://0.0.0.0:4443"

client = storage.Client(
    credentials=AnonymousCredentials(),
    project="test-project",
)

# initial bucket/file for docker image
bucket_name = "sample-bucket"
blob_name = "some_file.txt"

# test
bucket = client.get_bucket(bucket_name)
blob = bucket.blob(blob_name)
print(list(blob.acl))
blob.make_public() 
print(list(blob.acl))
blob.make_private()
print(list(blob.acl))

The outputs:
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}]
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}, {'entity': 'allUsers', 'role': 'READER'}]
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}]

The second snippet shows how for a newly created blob, the acl's also update when make_public() and make_private() are called:

os.environ["STORAGE_EMULATOR_HOST"] = "http://0.0.0.0:4443"

client = storage.Client(
    credentials=AnonymousCredentials(),
    project="test-project",
)

# buckets/files to create and test
upload_bucket_name = "test-bucket-with-globally-unique-name"
upload_blob_name = "test-blob-upload.svg"
upload_blob_file_path = "./image.svg"

# initialize
try:
    bucket = client.bucket(upload_bucket_name)
    bucket.storage_class = storage.constants.STANDARD_STORAGE_CLASS
    client.create_bucket(bucket, location="EU", retry=None)
except:
    pass
bucket = client.get_bucket(upload_bucket_name)
blob = bucket.blob(upload_blob_name)
blob.upload_from_filename(upload_blob_file_path, retry=None)

# test
bucket = client.get_bucket(upload_bucket_name)
blob = bucket.blob(upload_blob_name)
print(list(blob.acl))
blob.make_public()
print(list(blob.acl)) 
blob.make_private()
print(list(blob.acl))

The outputs:
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}]
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}, {'entity': 'allUsers', 'role': 'READER'}]
[{'entity': 'projectOwner-test-project', 'role': 'OWNER'}]

Explanation of my implementation:

For fixing #944:
Within main.go, when it reads all the existing files within the bucket, it previously never set an ACL for any of these objects. I assumed that for these pre-existing blobs, the ACL for each of them would just be [{'entity': 'projectOwner-test-project', 'role': 'OWNER'}].

For fixing #945:
From looking at how the python API updates the ACL's, I found that rather than sending a POST to the endpoint /b/{bucketName}/o/{objectName:.+}/acl, it instead sends a PATCH request to the endpoint /b/{bucketName}/o/{objectName:.+}.

Thus, I needed to update the patchObject method within fakestorage/object.go to detect if new ACL's are passed in and update the object's ACL if so.

[fsouza]

nit: can we call this payload?

Suggested change

	var dataInBody struct {
	var payload struct {

[fsouza]

Hmm this won't work with the backend filesystem, will it? We're modifying the value in memory and not persisting it. We need to update the PatchObject signature to take something like "attrsToUpdate" and apply changes to other attributes, not just metadata (this is necessary for #1024 too).

[RachitSharma2001]

Ah good point. I will fix this and push the changes.

[RachitSharma2001]

I have made the requested change. Let me know if what I did was best practice, and if there is any changes needed.

[fsouza]

This doesn't really need to be a global, let's inline it (I know it means we'll repeat it in tests, but that's OK).

[fsouza]

Thanks for working on this! Just one nit on generalizing some code, otherwise it's ready to go.

If you prefer, I can take care of it.

[fsouza]

Thank you very much for contributing!