Compressing the discovery cache about google-api-python-client

[parthea]

PRs are welcome!

We should ensure that when we retrieve the discovery artifact from the compressed file using get_static_doc that we only extract the file for the requested service and version.

google-api-python-client/googleapiclient/discovery_cache/__init__.py

Lines 55 to 78 in c965b05

	def get_static_doc(serviceName, version):
	"""Retrieves the discovery document from the directory defined in
	DISCOVERY_DOC_DIR corresponding to the serviceName and version provided.
	Args:
	serviceName: string, name of the service.
	version: string, the version of the service.
	Returns:
	A string containing the contents of the JSON discovery document,
	otherwise None if the JSON discovery document was not found.
	"""
	content = None
	doc_name = "{}.{}.json".format(serviceName, version)
	try:
	with open(os.path.join(DISCOVERY_DOC_DIR, doc_name), "r") as f:
	content = f.read()
	except FileNotFoundError:
	# File does not exist. Nothing to do here.
	pass
	return content

[abhinavsingh]

This also significantly impacts the container startup time (download/extract/run cycle). A better option can be to allow bundling of required json schema's only (instead of dumping every json spec in the container). Example, in our case we only use PubSub publisher client and GCS bucket APIs.

[sevdog]

How about putting compressed JSON schemas in the distributed package and have them to be decompressed and stored only after first usage? It could save space and be more efficient when requiring a specific API (since it will be decompressed and thus easier to read after first usage).