index user if not existed and log last active time for user accessing apis via backend apps

[ztsai]

closes #220

[coveralls]

Coverage decreased (-2.09%) to 85.947% when pulling c584d5b on backendUsers into c678c94 on master.

[MrOrz]

Commit 3 comments here first, still under review QQ

[MrOrz]

We may need to remove L8 to make this line work xd

[ztsai]

haha, good catch.
Actually, I think this might only be needed for tests to pass locally. It doesn't seem to timeout as easily on travis.
I am actually thinking, maybe we can make timeout value an environment variable to be set in .env and move this to setup since this is pretty much environment performance dependent (I don't know about everyone else, but they run way slower on my laptop than on travis :p)

[MrOrz]

I think it's OK that we increase timeout directly.

Increasing timeout does not impact time spent on test, but can save precious developing energy on problems that we can really solve.

[MrOrz]

Thanks a million for looping in tests for index.js! This is the first time apollo server context logic can be included in tests. I am grateful for @ztsai 's integration effort 🙇‍♀️

I am a bit confused how Apollo server context could interact with other middlewares like checkHeaders. We may need to check if the current logic works for web browser users.

Lastly, I have some nitpicking suggestions are regarding snapshot names :P

[ztsai]

frontend users are tested with logging in from development frontend, and backend users are tested with

curl "http://localhost:5000/graphql?userId=$USERID" -H 'Accept-Encoding: gzip, deflate, br' -H 'Content-Type: application/json' -H 'Accept: application/json' -H 'Connection: keep-alive' -H 'DNT: 1' -H 'Origin: http://localhost:5000' -H "x-app-secret:$SECRET" --data-binary '{"query":"query {\n  GetUser{\n  \tid\n  \tname\n    avatarUrl\n    level\n  }\n}"}' --compressed

[MrOrz]

LGTM! Thanks for the fix!

Added some nit-picking comments that should not be merge blockers. Feel free to adapt and merge anytime :)

[MrOrz]

Thanks for the refactor! Let's merge this. I will try deploy it to staging before wednesday, along with data migration.