feat(types): allow key separator augmentation

[ImADrafter]

ATM, type checking the dictionaries will not work properly if you have changed the keySeparator for javascript:

i18n.init({
  ...
  keySeparator: '>>>',
)}:

t('key>>>nested') // Will translate at runtime, but typescript will complain, as he is always expecting the default key separator. 

The changes that im proposing allows to augment the keySeparator, in order to maintain aligned the typescript interface and the i18n instance.

Nevertheless, im not happy with all the code changes cause i had to rewrite NormalizeReturn, and it became too complex to really understand all the chaining conditions, so if you have any clue of how it should remain simple, please let me know.

Thank you in advance for reviewing this changes.

Checklist

• only relevant code is changed (make a diff before you submit the PR)
• run tests npm run test
• tests are included

Checklist (for documentation change)

• only relevant documentation part is changed (make a diff before you submit the PR)
• motivation/reason is provided

[coveralls]

Coverage remained the same at 96.416% when pulling 9109e04 on ImADrafter:feat-augmeted-key-separator into 04b2bca on i18next:master.

[pedrodurek]

Seems great to me, I'll just do some performance tests tomorrow to ensure compilation time won't be jeopardized.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[pedrodurek]

Hey sorry for the delayed response, packing and unpacking when moving provinces is not easy 😅 . PR is good to go!

Just for the record:
This PR is slowing down the compilation time by ~23% when inferring the keys for the first time (big projects).

Before:

After:

[ImADrafter]

Worry not @pedrodurek , thank you for taking the time after all :)

Im a little bit disappointed to see that compilation time increases, I think it shouldn't... Do you have any idea on how can we improve performance or can we move foward?

BTW, how can I test myself performance? are you using a project that you can share?

[pedrodurek]

Hey @ImADrafter don't worry, it didn't increase much. That's normal, the more validation we add, the more time it'll take since it iterates over every key doing the same check.

I'm testing using this repo https://github.com/pedrodurek/i18n-type-tests, to test performance you need to enable the Ts Server log, you can find more info here.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[adrai]

so ready to merge?

[pedrodurek]

yep!

[adrai]

included in v11.13.0