feat(no-unlocalized-strings): remove default configuration

[timofei-iatsenko]

Quote myself:

I went through the issues with this rule and tried to address them all. It seems like we need a big change in how the rule is set up. There’s no way to make it work perfectly for everyone and every project. So, I’ve decided to remove all the built-in settings and made them fully customizable.

For example, if a user doesn’t want:

but we noticed that const CAPITAL_LETTERS is always ignored(?).

they can simply remove the related regex from their options.

This will fix

• Support Chinese Please #28 (chinese users will be able to set their own regexp)
• const CAPITAL_LETTERS is always ignored

And possible many more issues in the future regarding the defaults.

What was done:

1. All hard-coded default are removed in favor of user configuration
2. Main words regex was extended to support non-Latin letters, such as Chinese or Cyrillic.
3. ignoreAttribute, ignoreProperty and ignoreVariable options were merged into single ignoreNames
4. strictAttribute option removed. It didn't work properly, and now it's not needed because the user can override defaults more naturally (that was an original request for this option).
5. rule is not triggered for lingui's msg and selectOrdinal functions

Breaking Changes

1. The rule will not have any defaults. You need to set defaults in eslint config for this rule. Refer to rule documentation for the basic configuration.
2. The ignoreFunction option was renamed to ignoreFunctions
3. ignoreAttribute, ignoreProperty and ignoreVariable options were merged into single ignoreNames
4. strictAttribute option removed.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.49%. Comparing base (e249254) to head (0bfae9b).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #78      +/-   ##
==========================================
- Coverage   96.61%   96.49%   -0.12%     
==========================================
  Files          10       10              
  Lines         413      371      -42     
  Branches      118      101      -17     
==========================================
- Hits          399      358      -41     
+ Misses         14       13       -1     

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

[jgarplind]

I can see why you want to go in this direction, but personally I had a lot of help from the defaults, and might have been hesitant to even try to use the rule if faced with a custom setup.

UPPERCASE_LETTERS being ignored was fine imo. My 10/10 experience would be what we had, including the ability to override such specifics if desired (unsure how feasible that is).

Either way, I'm a bit wary here, thinking that this PR may be a move away from a 9/10 out of the box experience which frankly is pretty good.

[timofei-iatsenko]

The problem is a could not come up with a good way to override defaults. So we have two situations:

1. When user ok with provided defaults and want to add more specificity for the project
2. User is not ok with defaults and want to revert some of the buil-in rules (that already happened few times, saying from issues)

First case is already supported, we can extend the defaults.

The second case is a bit trickier, you need to give a way for user to choose what is going to happen, override or extend? This either brings more complicated structure of every option or double of options (ignorePoperty, ignorePropertyOverride). I don't like both of variants, actually.

That's why I decided to give a blank list, so there is nothing to override and it's pretty configurable. Also, it gives much more clarity about what exactly is ignored

[jgarplind]

Solid reasoning.

Your example is thorough and should be easy enough to copy paste and run with. I might have been further encouraged if this language

You can use the following config as a starting point and then adjust it for your project

was slightly more assertive, i.e. officially recommending it, or mentioning it's history as the de facto-rule set before opening up for configuration.

[timofei-iatsenko]

@andrii-bodnar this one is ready to be merged.

[andrii-bodnar]

@timofei-iatsenko thank you!