fix(reports): stable total order on SIE-export fetchAllRows paging (#793 hardening) (#824)

#820 fixed the primary #793 truncation (the nested embedded-resource select hit
PostgREST's row ceiling, exporting ~30 vouchers) by splitting entries and lines
into two fetchAllRows calls. But both queries paged without a stable TOTAL order:
the entries query ordered by voucher_number alone (not unique across voucher
series) and the lines query had no .order() at all. Per the fetch-all.ts
invariant, .range() paging without a unique total order can duplicate or skip
rows across the 1000-row boundary — so a year with >1000 entries/lines and
multiple series could still drop or double a voucher in the SIE file (BFL
completeness).

- Entries: order by voucher_series + voucher_number (unique per company+period).
- Lines: order by the line PK id.
- Both carry dedupeBy: r => r.id as defense-in-depth, mirroring the general
  ledger / trial balance fix in #811.
- Give the large-period test's line fixtures unique ids so the >1-page dedupe
  path is exercised realistically (all 5000 lines survive).

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

lib/reports/__tests__/sie-export.test.ts

@@ -374,9 +374,9 @@ describe('generateSIEExport', () => {
       status: 'posted',
     }))
 
-    const lines = entries.flatMap((e) => [
-      { journal_entry_id: e.id, account_number: '1510', debit_amount: 100, credit_amount: 0, line_description: null, cost_center: null, project: null },
-      { journal_entry_id: e.id, account_number: '3001', debit_amount: 0, credit_amount: 100, line_description: null, cost_center: null, project: null },
+    const lines = entries.flatMap((e, i) => [
+      { id: `l${i * 2 + 1}`, journal_entry_id: e.id, account_number: '1510', debit_amount: 100, credit_amount: 0, line_description: null, cost_center: null, project: null },
+      { id: `l${i * 2 + 2}`, journal_entry_id: e.id, account_number: '3001', debit_amount: 0, credit_amount: 100, line_description: null, cost_center: null, project: null },
     ])
 
     // fetchAllRows paginates at PAGE_SIZE = 1000; chunk the mock data so the

lib/reports/sie-export.ts

@@ -74,8 +74,15 @@ export async function generateSIEExport(
       q = q.neq('source_type', 'year_end')
     }
 
-    return q.order('voucher_number').range(from, to)
-  })
+    // Stable TOTAL order: voucher_series + voucher_number is unique per
+    // company+period, so fetchAllRows paging can't duplicate or skip a voucher
+    // across the 1000-row boundary on large years (voucher_number alone is not
+    // unique across series). dedupeBy is defense-in-depth — see fetch-all.ts.
+    return q
+      .order('voucher_series', { ascending: true })
+      .order('voucher_number', { ascending: true })
+      .range(from, to)
+  }, { dedupeBy: (r) => r.id })
 
   // Fetch all lines for those entries, filtered server-side via an inner join
   // so the same company/period/status (and year-end exclusion) constraints
@@ -92,9 +99,11 @@ export async function generateSIEExport(
       q = q.neq('journal_entries.source_type', 'year_end')
     }
 
+    // Stable total order on the line PK so paging can't duplicate/skip a line
+    // across the 1000-row boundary; dedupeBy is the defense-in-depth net.
     // eslint-disable-next-line @typescript-eslint/no-explicit-any
-    return q.range(from, to) as any
-  })
+    return q.order('id', { ascending: true }).range(from, to) as any
+  }, { dedupeBy: (r) => r.id })
 
   const linesByEntryId = new Map<string, JournalEntryLine[]>()
   for (const line of allLines) {