Major overhaul

[sorbits]

Add ruby version of workflow

I realize that it is not popular to change the implementation to another language, but I felt a major refactoring was called for to continue adding desired features, and as I work a lot faster in ruby, I decided to do it in ruby.

If you do accept this PR then I suggest that you make me a collaborator of this repository.

One thing that this rewrite does not handle is auto-update, but since you’re not using GitHub Releases, it would appear that you were not actually making use of the alfred-workflow auto-update support.

I think there should be an auto-update feature, I can add this once I know whether or not you agree to make me a collaborator, as I think we should use GitHub Releases for updates.

Some of the changes/improvements that this rewrite brings:

• When we have cached data, we immediately return a response to Alfred based on the cache. If the cace is older than 10 minutes, we refresh the cache in a background process and instruct Alfred to reload the list again after 0.8 seconds.
• Removed d > Clear Cache and d > Set Cache [seconds] because of the improved caching behavior.
• The create actions accept an optional document name and are no longer under d >, for example d ns Budget should now create a spreadsheet named “Budget”. Closes #11.
• Removed d > Login as we automatically open the Google Sign In page when refresh token is missing/invalid.
• Changed d > Logout to d Sign out of Google Drive and made it perform an account sign out (instead of just deleting the access tokens).
• Refresh access token when expiration time is less than 10 seconds into the future. This avoids an extra (failing) server request/response, which is mainly for the document creation actions, as Google’s servers are on the slow side.
• The OAuth2 HTTP callback server is running in our main process, this means that we can perform actions that require authentication after we receive the callback from Google, for example d new document without a valid token will now first ask the user to sign in and then create the new document. It also improves robustness slightly because error handling is in the same process.
• Cache is refreshed (in the background) after creating a new document.
• Add verbose (timestamped) logging which should make troubleshooting easier.
• The OAuth2 HTTP callback server will run until we see the expected GET / request (so not fooled by a favicon.ico request or similar) and it uses a URL parser to obtain the code query parameter from the callback URL.

The above should fix #5, #16, #23, and #25.

You can also close #28 (but I already did that in 6006836b37aaef9335ca4f48eb87bb4beda27a23).

[azai91]

@sorbits would you like to be added as a contributor to this project?

[sorbits]

would you like to be added as a contributor to this project?

Yes, I think that would be good for both of us, thanks!