fix(nexa_coa_setup): clean GL codes — 119100/119900->115200/115900, 511105->511100

Three odd code positions that were chosen to dodge l10n_ca collisions are
now cleaned up:
- Due From Shareholder       119100 -> 115200 (115xxx is where receivables belong)
- Due From Associated Corps  119900 -> 115900
- Cloud Infrastructure       511105 -> 511100 (legacy 'Inside Purchases'
                                       renamed to 511100.OLD)

Applied to prod via scripts/fix_gl_codes.py (now committed).

Module XML updated: <field name='code'> values match new codes; XMLIDs
(acct_119100, acct_119900, acct_511105) preserved so existing
ir.model.data rows on prod still map to the right records.

pre_init_hook augmented with _L10N_CA_FORCE_CLEAR_CODES set so a fresh
install on a new DB also force-clears 511100 (which would otherwise be
blocked by the postings-exist guard).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

nexa_coa_setup/data/01_account_account.xml

@@ -8,16 +8,20 @@
 
         <!-- NOTE: 115100 is l10n_ca "Customers Account" (240 postings — AR control) — kept as l10n_ca.
              115110 is l10n_ca "Customers Account (PoS)" — kept.
-             Nexa intercompany receivables live in the 119xxx range to avoid all collisions. -->
+             Nexa intercompany receivables sit at 115200 / 115900 (freed when their
+             unused l10n_ca defaults — Mortgage Loans — were deleted during install).
+             XMLIDs (acct_119100, acct_119900) preserved from initial install where
+             these accounts were at codes 119100/119900 — codes updated in-place via
+             scripts/fix_gl_codes.py without rewriting ir.model.data. -->
         <record id="acct_119100" model="account.account">
-            <field name="code">119100</field>
+            <field name="code">115200</field>
             <field name="name">Due From Shareholder — Gurpreet</field>
             <field name="account_type">asset_current</field>
             <field name="reconcile" eval="True"/>
         </record>
 
         <record id="acct_119900" model="account.account">
-            <field name="code">119900</field>
+            <field name="code">115900</field>
             <field name="name">Due From Associated Corporations</field>
             <field name="account_type">asset_current</field>
             <field name="reconcile" eval="True"/>
@@ -348,10 +352,12 @@
         <!-- 5xxxxx — DIRECT COSTS (COGS)                                 -->
         <!-- ============================================================ -->
 
-        <!-- NOTE: 511100 was "Inside Purchases" in l10n_ca (1 posting from legacy bookkeeping) — kept as l10n_ca.
+        <!-- NOTE: legacy l10n_ca "Inside Purchases" at 511100 (had 1 historical
-             Cloud Infrastructure sits at 511105 to avoid collision. -->
+             posting) was renamed to 511100.OLD via scripts/fix_gl_codes.py —
+             Cloud Infrastructure now claims the clean 511100 code. XMLID
+             acct_511105 preserved from initial install. -->
         <record id="acct_511105" model="account.account">
-            <field name="code">511105</field>
+            <field name="code">511100</field>
             <field name="name">Cloud Infrastructure (AWS, Hetzner, OVH, DigitalOcean, Linode)</field>
             <field name="account_type">expense_direct_cost</field>
         </record>

nexa_coa_setup/hooks.py

@@ -4,14 +4,12 @@ import logging
 _logger = logging.getLogger(__name__)
 
 
-# l10n_ca account codes that collide with the Nexa CoA design and that
+# l10n_ca account codes that collide with the Nexa CoA design. Each is
-# l10n_ca pre-loads with 'income_other'/'expense'/etc. types we don't want.
+# checked at pre_init: if it has zero postings we suffix its code with
-# Each of these is checked at pre_init: if it has zero postings we suffix
+# '.OLD' and archive it so our XML can claim the code.
-# its code with '.OLD' and archive it so our XML can claim the code.
+# Codes with postings are LEFT ALONE — we renumbered the Nexa account.
-# Codes with postings are LEFT ALONE — we renumbered the Nexa code instead
+# Currently 115100 stays as l10n_ca 'Customers Account' (240 postings, AR
-# (115100 stays as l10n_ca 'Customers Account' AR; Nexa shareholder receivable
+# control) — Nexa shareholder receivable sits at 115200 instead.
-# moved to 119100. 511100 stays as l10n_ca 'Inside Purchases'; Nexa Cloud
-# Infrastructure moved to 511105).
 _L10N_CA_COLLISION_CODES = [
     "118100", "118200", "118300",
     "213100", "214100",
@@ -19,12 +17,18 @@ _L10N_CA_COLLISION_CODES = [
     "311100", "311200", "311300",
     "411100", "411200", "411300",
     "413100", "413200", "413300",
-    "511110", "511120", "511130", "511140", "511200", "511210",
+    "511100", "511110", "511120", "511130", "511140", "511200", "511210",
     "512100", "512110", "512200",
     "611100", "611200", "611300",
     "612100", "612200",
 ]
 
+# Codes that MUST be cleared even if they have postings (force-suffix to .OLD).
+# Use sparingly — historical reports lose the original name. Only for codes
+# where the Nexa account at that code is the canonical one going forward and
+# any prior posting is a misclassification the user will re-class later.
+_L10N_CA_FORCE_CLEAR_CODES = {"511100"}
+
 
 def pre_init_hook(env):
     """Run BEFORE XML data is loaded. Clear l10n_ca account codes that would
@@ -47,7 +51,8 @@ def _clear_l10n_ca_collisions(env):
             not_found += 1
             continue
         usage = env["account.move.line"].search_count([("account_id", "=", acc.id)])
-        if usage > 0:
+        force = code in _L10N_CA_FORCE_CLEAR_CODES
+        if usage > 0 and not force:
             _logger.info(
                 "nexa_coa_setup: keeping l10n_ca account %s (%s) — %d postings exist",
                 code, acc.name, usage,
@@ -64,6 +69,11 @@ def _clear_l10n_ca_collisions(env):
             "active": False,
         })
         cleared += 1
+        if force and usage > 0:
+            _logger.info(
+                "nexa_coa_setup: force-cleared %s despite %d postings (in FORCE_CLEAR set)",
+                code, usage,
+            )
     _logger.info(
         "nexa_coa_setup: collision sweep — cleared %d, kept-with-postings %d, not-found %d",
         cleared, kept_with_postings, not_found,

nexa_coa_setup/scripts/fix_gl_codes.py

@@ -0,0 +1,40 @@
+"""Renumber a few oddly-coded accounts to cleaner positions.
+Idempotent — checks before each move."""
+moves = [
+    # (current_code, target_code, expected_name_substr)
+    ('119100', '115200', 'Due From Shareholder'),
+    ('119900', '115900', 'Due From Associated Corporations'),
+]
+print(">>> Easy renames (no postings)")
+for old, new, name_hint in moves:
+    acc = env['account.account'].search([('code', '=', old)], limit=1)
+    if not acc:
+        print(f"SKIP {old}: not found")
+        continue
+    if name_hint.lower() not in (acc.name or '').lower():
+        print(f"SKIP {old}: name doesn't match expected '{name_hint}' (got '{acc.name}')")
+        continue
+    conflict = env['account.account'].with_context(active_test=False).search([('code', '=', new)], limit=1)
+    if conflict:
+        print(f"SKIP {old}->{new}: target code occupied by {conflict.name}")
+        continue
+    acc.code = new
+    print(f"OK {old} -> {new}: {acc.name}")
+
+# The 511100 swap — rename legacy first, then renumber ours
+print(">>> 511105 -> 511100 swap")
+legacy = env['account.account'].with_context(active_test=False).search([('code', '=', '511100'), ('name', 'ilike', 'inside purchases')], limit=1)
+ours = env['account.account'].search([('code', '=', '511105'), ('name', 'ilike', 'cloud infrastructure')], limit=1)
+if not legacy:
+    print("legacy 511100 not found (already moved?)")
+elif not ours:
+    print("our 511105 not found (already renamed?)")
+else:
+    # Rename legacy first to free the code
+    legacy.write({'code': '511100.OLD', 'name': f"(l10n_ca LEGACY) {legacy.name}", 'active': False})
+    ours.code = '511100'
+    print(f"OK legacy 511100 -> 511100.OLD ({legacy.name})")
+    print(f"OK 511105 -> 511100 ({ours.name})")
+
+env.cr.commit()
+print(">>> done <<<")