[Aleksandr Gladkov (AleksanderGladkov)]

Backport of #12696 for AB#653927

[alexei-dobriansky]

Good Sense Reviewer - Round 1

Recommendation: Accept

What this PR does

This change recalculates SAF-T 1.30 debit and credit control totals from the sign of each selected G/L entry amount. It keeps SAF-T 1.20 behavior unchanged and repeats the full selection totals in every split file.

The new aggregate uses the same posting-date and source-code filters as the line exporter. Positive amounts are summed as debits, while negative amounts are converted to positive credit totals, which matches the XML line classification. The regression tests cover reversals, negative entries, multiple and blank source codes, period limits, empty selections, foreign currency, and monthly and daily splits.

Problem-solution fit

Fit: Strong

The change directly fixes the reported mismatch between the control totals and the exported line amounts. Its scope is limited to SAF-T 1.30, where the XML uses sign-based debit and credit classification.

Suggestions

None.

Risk assessment and necessity

Risk: This affects financial audit-file control totals, so a wrong sign or filter could produce a non-compliant export. The implementation reuses the exporter's source-code selection, uses a read-only SQL aggregate, changes no public API or event contract, and has focused regression coverage.

Necessity: Without this change, reversals and other negative entries can make SAF-T 1.30 control totals disagree with the amounts in the exported files. The fix is targeted and preserves the existing SAF-T 1.20 calculation.

---

[AI-PR-REVIEW] version=1 promptVersion=4 system=github pr=12899 round=1 by=alexei-dobriansky at=2026-10-15T14:12:54Z lastSha=f1bc370ecd1f1597ab7e5d9c688973db4f06e0a6 reviewKey=73722dfbecf39ccd0956cd08e6d9a0d47efaa8a0736b4dc335f28bf55050c297 suggestions=