fix: force passed locales into faker constructor

[vith]

I saw that there's a Faker class being exported now, so I tried to use it, but it doesn't seem to work.

For example, when trying to do faker = new Faker(); faker.name.firstName() I get:

TypeError: Cannot read properties of undefined (reading 'name')

I didn't see any examples of using this constructor in the docs, so it's possible I'm using it wrong. Anyway, I wrote these tests to check if it was a problem in my own project, but they fail as well in the same way.

[Shinigami92]

I think it fails, because you don't provide any locales.

Maybe we need to change the type on your side to force to provide a locales 🤔

[Shinigami92]

Uhm... tests are green? 👀

[codecov]

Codecov Report

Merging #580 (fed2179) into main (814880b) will increase coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head fed2179 differs from pull request most recent head 1870549. Consider uploading reports for the commit 1870549 to get more accurate results

@@           Coverage Diff           @@
##             main     #580   +/-   ##
=======================================
  Coverage   99.34%   99.34%           
=======================================
  Files        1923     1923           
  Lines      176891   176903   +12     
  Branches      896      901    +5     
=======================================
+ Hits       175735   175748   +13     
+ Misses       1100     1099    -1     
  Partials       56       56           

Impacted Files	Coverage Δ
src/faker.ts	100.00% <100.00%> (ø)
src/index.ts	97.43% <0.00%> (+2.56%)	⬆️

[ST-DDT]

I assume since it probably returns undefined in the middle the fallback of ['a', 'b', 'c'] applies and thus passes this trivial test.
Maybe change it to length > 1 (as there shouldn't be any one char names in our data).

See also: #581

[vith]

The tests are green just because I defined them with it.fails. If you want I can remove that so they actually cause the suite to fail now.

I see what you mean now about not providing any locales, i.e. I need replicate something like this:

import { Faker } from '../faker';
import sv from '../locales/sv';
import en from '../locales/en';

const faker = new Faker({
  locale: 'sv',
  localeFallback: 'en',
  locales: {
    sv,
    en,
  },
});

export = faker;

But I haven't found any way to actually import the locale data externally. I think the exports field of package.json only lets me import the pre-instantiated per-locale singletons like the one I pasted above from src/locale/sv.ts?

[ST-DDT]

The tests are green just because I defined them with it.fails. If you want I can remove that so they actually cause the suite to fail now.

Yes, that would make it more obvious for anyone looking at the tests.

But I haven't found any way to actually import the locale data externally. I think the exports field of package.json only lets me import the pre-instantiated per-locale singletons like the one I pasted above from src/locale/sv.ts?

Yes, currently there isn't because it wasn't in 5.x. This is planned for a release after 6.0 when we actually start adding fixes and features.

[vith]

Yes, that would make it more obvious for anyone looking at the tests.

Done.

Yes, currently there isn't because it wasn't in 5.x. This is planned for a release after 6.0 when we actually start adding fixes and features.

Ah, okay. I wonder if the Faker class shouldn't be exported from the package in 6.0 then, if the locale data it needs won't be?

I tried to use the class because I got it as a code completion via the TypeScript Language Server in VSCode.

Not sure if there's anything useful to make out of the tests in this PR then. I guess making the locales field required in the constructor would help like @Shinigami92 said, and/or a better error message when locale data is missing.

[ST-DDT]

Could you adjust the constructor to not have a default and require the locales?

[Shinigami92]

Could you adjust the constructor to not have a default and require the locales?

@vith Would you like to open a new PR to do this or maybe just use this PR?

[ST-DDT]

Could you adjust the constructor to not have a default and require the locales?

@vith Would you like to open a new PR to do this or maybe just use this PR?

I recommend using this one, as both parts fix+test belong together anyway.

[Shinigami92]

I will jump into this PR now and fix the constructor

[ST-DDT]

This needs to be rebased