[19.0][MIG] payroll_account: Migration to 19.0

[neoand]

Migration of payroll_account from 18.0 to 19.0.

Module history was replayed with git format-patch --keep-subject --stdout origin/19.0..origin/18.0 -- payroll_account | git am -3 (69 commits), and the migration itself is the single commit on top.

What Odoo 19 broke

Odoo 19 removed the hr_contract addon: contract data now lives on hr.version, which is exactly the model this module extends. The payroll module already made that jump on 19.0, so this migration follows the same pattern.

Change	Fix
hr_contract addon is gone	models/hr_contract.py: _inherit "hr.contract" → "hr.version"
Parent view hr_contract.hr_contract_view_form no longer exists	Inherit payroll.hr_contract_view_form (model hr.version) and anchor the accounting fields after struct_id
hr.employee.gender → sex	Fixed in tests
hr.employee.bank_account_id (m2o) → bank_account_ids	Fixed in tests

Tests now build the contract through employee.version_id.write() with contract_date_start, the pattern used by payroll/tests/common.py on 19.0.

The 18.0.1.0.2 migration scripts were dropped.

Testing

On odoo:19.0 Community:

Module payroll_account: 0 failed, 0 error(s) of 3 tests

Beyond the module's own suite, I checked the feature end-to-end on a fresh database: a payslip whose rules carry account_debit / account_credit, confirmed with action_payslip_done(), produces a balanced account.move posted on the journal configured on the version.

Entry MISC/2026/04/0001 | journal=Miscellaneous Operations | state=posted
ACCOUNT   LABEL          DEBIT       CREDIT
443000    Salary     12,000.00         0.00
201000    Salary          0.00    12,000.00
201000    ISR             0.00     1,200.00
443000    ISR         1,200.00         0.00
          TOTALS     13,200.00    13,200.00

I also re-ran a payslip-engine check on payroll alone before and after installing this module to confirm nothing regressed.

Note

Loading demo data changed in Odoo 19 — --without-demo is no longer enough and there is a new odoo module force-demo subcommand. Without demo data these tests fail on base.res_partner_12, which does exist in 19.0's demo. Worth knowing when reproducing locally.

@appstogrow @nimarosa

---

Assisted-by: Claude Opus 4.8

Disclosed per the OCA Generative AI / LLM Policy.
Developed and verified under my direction and review; I take responsibility for the contribution.

[stferraro]

Hi @neoand,

I noticed this PR came in after mine (#269, opened back in May), so I wanted to flag something before they diverge further.

While working on #269 I found and fixed a real bug in payroll_account's journal handling: when generating payslips in bulk from a hr.payslip.run batch, the batch's journal was silently overriding the employee's own contract journal (hr.version.journal_id) instead of respecting it as a fallback. Worth checking if your migration carries the same behavior, since the original v18 code has it too.

Changes in #269 that address this:

create() override in hr.payslip now gives priority to the contract's own journal_id, falling back to the batch's journal only when the contract has none.
Added journal_id to _get_whitelist_fields_from_template() so it's actually copied when a contract template is applied — it wasn't before.
Test coverage for both the standalone create() path and the batch wizard, plus the account-move balancing paths that weren't covered before.
Since #269 predates yours and already has this fix in place with tests, would you mind taking a look and reviewing it? Happy to also look at yours and see if there's anything worth merging between the two. No rush — just wanted to make sure this doesn't get lost with two open migrations for the same module.

Thanks for your work on this either way!

[neoand]

Cierro este PR por decisión interna de pausar contribuciones a OCA a partir del 15-ago-2026.

El código queda archivado en mi fork privado (neoand/arw-ivan) para uso local del proyecto de migración ARW.

Agradezco especialmente a @stferraro por avisar a tiempo que existía #269, y por la conversación que se abrió. El trabajo de #269 cubre esta migración con tres aprobaciones, CI en verde, y el fix del bug de journal_id en el template de contratos. Cerrar este a favor del suyo era lo correcto.

Gracias a los mantenedores por su tiempo en revisarlo.

— Anderson