[MOD][18.0] hr_employee_language: languages selection

[mymage]

As in #1486
Changes the source of language list moving it from the res.lang.csv file to res.lang model.
In this way if you add a language at runtime it will be immediately available.

Added _compute_display_name (ex name_get) to show language name instead language code on selection or m2o widget

[mymage]

@dreispt @primes2h @TheMule71 plase review.
Little differences from #1512

[mymage]

@dreispt Thanks for your notes; I'm not so skilled on coding so your suggestion is very appreciated. I'll give a look asap.

[mymage]

@dreispt Thanks to your comments I refactored the code. The only "strange" thing that I made is this:
self_data = self.env["hr.employee.language"].search([]) for record in self_data:
to have CodeCov at 100%: this because despite I create an element in hr.employee.language for test, here self is empty.

[mymage]

@dreispt @primes2h @TheMule71 ping. Thanks

[mymage]

@dreispt @primes2h @TheMule71 ping. Thanks

[alexey-pelykh]

Thanks for improving the language selection to use res.lang records. A couple of things to address:

1. _compute_display_name() searches ALL hr.employee.language records (self.env["hr.employee.language"].search([])) instead of iterating over self. This is both a performance issue and semantically incorrect — compute methods should only set values on self records.

2. The compute method returns a list of tuples, but Odoo compute methods don't use return values — they assign to self.display_name (which you do, but the return is dead code).

3. Using _() on selection keys (_("%(code)s", code=lang.code)) translates the code itself, which shouldn't be translated — codes are stable identifiers.

Happy to re-review once updated.

[mymage]

Thanks @alexey-pelykh for your suggestions. I updated the code, I hope in the right way.
If this is ok I will fix the other two.

[alexey-pelykh]

Thanks for the update — the model code now correctly addresses all three issues from my previous review:

• _compute_display_name iterates self instead of searching all records
• No more dead return value
• No more _() on language codes

However, the test for _compute_display_name doesn't actually exercise the compute method:

self.env["hr.employee.language"].create({...})  # result not stored
self.env["hr.employee.language"]._compute_display_name()  # called on empty recordset

Since the create result isn't assigned, _compute_display_name() runs on an empty recordset — the loop body never executes. This is likely why codecov is failing. The fix:

record = self.env["hr.employee.language"].create({
    "name": "en_GB",
    "employee_id": self.env["hr.employee"].create({"name": "Test"}).id,
})
self.assertEqual(record.display_name, "English (UK)")

Accessing record.display_name triggers the compute automatically — no need to call it explicitly.

Co-Reviewed-By: Claude Opus 4.6 (1M context) noreply@anthropic.com

[mymage]

Ok @alexey-pelykh another thanks for your suggestions. Now I think it is ok but I don't understand the last red codecov

[alexey-pelykh]

Thanks for the update @mymage — the test now correctly exercises the compute method. All feedback from both reviews is addressed.

Regarding the codecov/project failure: this check measures overall repository coverage, not your PR's coverage. Your codecov/patch check passes, meaning your new code is adequately covered. The codecov/project failure is a pre-existing coverage gap in the repo — not something you need to fix here.

Co-Reviewed-By: Claude Opus 4.6 (1M context) noreply@anthropic.com