Stop using rawState in exportDataEnvelope

[ifosli]

📲 What

This is a cleaner way to handle any future server changes; I just didn't understand decoders well enough to to this initially. The actual behavior (unknown strings default to .unknown and anything that can't be parsed as a string throws an error) remains the same.

✅ Acceptance criteria

• Tests pass
• Project builds & personal data can still be requested

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (5e5cae9) 83.73% compared to head (9c7c6aa) 83.75%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1903      +/-   ##
==========================================
+ Coverage   83.73%   83.75%   +0.01%     
==========================================
  Files        1228     1229       +1     
  Lines      111837   111842       +5     
  Branches    29748    29749       +1     
==========================================
+ Hits        93649    93670      +21     
+ Misses      17163    17147      -16     
  Partials     1025     1025              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[amy-at-kickstarter]

This LGTM, but I'd like to see a really quickie test just for a valid and an invalid enum value.

I'm also slightly surprised that the default implementation of Decodable for an Enum doesn't already do this - what happened before you implemented this change?

[ifosli]

This LGTM, but I'd like to see a really quickie test just for a valid and an invalid enum value.

I'm also slightly surprised that the default implementation of Decodable for an Enum doesn't already do this - what happened before you implemented this change?

If it's just decoded directly (and there's no enum that matches the string found), it throws an error, which means the entire export data envelope doesn't get parsed. (That was the original cause of one of my starter bugs and it caused the "request my data" button to never be clickable.)

I didn't create tests for this initially because this file doesn't have any tests, but good call! Added now.

[amy-at-kickstarter]

This LGTM, but I'd like to see a really quickie test just for a valid and an invalid enum value.
I'm also slightly surprised that the default implementation of Decodable for an Enum doesn't already do this - what happened before you implemented this change?

If it's just decoded directly (and there's no enum that matches the string found), it throws an error, which means the entire export data envelope doesn't get parsed. (That was the original cause of one of my starter bugs and it caused the "request my data" button to never be clickable.)

I didn't create tests for this initially because this file doesn't have any tests, but good call! Added now.

Ah, gotcha, so this is replacing an unsafe/exploding init with a safe one. That seems fine for this use case!

[amy-at-kickstarter]

LGTM