Image Gallery Demo
This is a small demo of how to refactor a web application from using local file storage to using Cloud Storage (in this case, Google Cloud Storage buckets, but the principle applies to any cloud vendor object storage service.)
Why
In years of moving applications to the cloud, I am still surprised to see lots of code that relies on local POSIX filesystems. This makes it hard to take advantage of cloud-native reliability without spinning up some sort of redundant file server. It's a complete waste of time when off-the-shelf object storage is already available, extremely cheap to use, and already 10x more reliable than any custom solution you can come up with.
The original app in 1-local-storage relies on writing to a local filesystem. This is because the app is silly and thinks it is running on a server.
The refactored app in 2-cloud-storage uses GCS for object storage. Now the app itself is stateless and can run anywhere and in any configuration (containerised, auto-scaled etc.).
The version in 3-firestore adds support for storing image metadata in Cloud Firestore.
The version in 4-secrets is a demonstration of using sensitive data (in this case, an API key) in Cloud Run. The app expects an API_KEY variable. This can be injected into the runtime environment either manually with gcloud deploy run or via Cloud Build.
Note: For the sake of clarity there is no exception handling in this code. It is for demonstration purposes only. Use it at your own risk!