fix: normalize Factur-X preview data

Florian Wunderlich · Florian Wunderlich · commit 68051b4d28db · 2026-07-20T17:12:55.000+02:00
diff --git a/models/ai_auto_apply.py b/models/ai_auto_apply.py
@@ -23,11 +23,14 @@ class AccountMove(models.Model):
     def _ai_can_auto_apply(self, data):
         """Check if extraction can be auto-applied (skip preview).
 
-        Conditions: enabled in settings, vendor matched + reliable,
-        all confidence scores >= threshold, no warnings, valid doc type.
+        Conditions: enabled in settings, not explicitly preview-only,
+        vendor matched + reliable, all confidence scores >= threshold,
+        no warnings, valid doc type.
         """
         if not self._ai_get_bool_param('ai_auto_apply_enabled'):
             return False
+        if data.get('_require_preview'):
+            return False
 
         # Quick disqualifiers
         doc_type = data.get('document_type', '')
diff --git a/models/ai_extraction_engine.py b/models/ai_extraction_engine.py
@@ -98,6 +98,18 @@ def _ai_trigger_extraction(self, api_key, attachment, preview=False):
         data = self._ai_run_pipeline(api_key, cfg, raw_data, mimetype)
         if data is None:
             return
+        if preview and isinstance(data, dict) and data.get('_facturx'):
+            # Preview/cache/UI code expects the normal extraction schema, not
+            # the internal Factur-X wrapper with raw XML bytes. Keep explicit
+            # metadata so preview-mode Factur-X results cannot bypass review.
+            parsed_data = self._ai_parse_facturx(data['_xml'])
+            if parsed_data is None:
+                return
+            data = {
+                **parsed_data,
+                '_source': 'facturx',
+                '_require_preview': True,
+            }
         if preview:
             return data
         if isinstance(data, dict) and data.get('_facturx'):
diff --git a/models/ai_field_mapper.py b/models/ai_field_mapper.py
@@ -39,19 +39,26 @@ def _ai_get_invoice_attachment(self):
     # Factur-X application
     # ===================================================================
 
+    def _ai_parse_facturx(self, xml_data):
+        """Parse Factur-X XML into the standard extraction dict."""
+        self.ensure_one()
+        try:
+            return ai_facturx_parser.parse_facturx_xml(xml_data)
+        except Exception as exc:
+            _logger.warning('Factur-X XML parsing failed: %s', exc)
+            self.ai_extraction_status = 'failed'
+            self.ai_confidence = json.dumps({'source': 'facturx', 'overall': 0.0})
+            return None
+
     def _ai_apply_facturx(self, xml_data):
         """Apply Factur-X structured data directly to the invoice (zero AI cost).
 
         Parses the CII XML into the same dict format as Claude's response,
         then feeds it through the standard extraction pipeline.
         """
         self.ensure_one()
-        try:
-            data = ai_facturx_parser.parse_facturx_xml(xml_data)
-        except Exception as exc:
-            _logger.warning('Factur-X XML parsing failed: %s', exc)
-            self.ai_extraction_status = 'failed'
-            self.ai_confidence = json.dumps({'source': 'facturx', 'overall': 0.0})
+        data = self._ai_parse_facturx(xml_data)
+        if data is None:
             return
 
         try:
diff --git a/tests/test_auto_apply.py b/tests/test_auto_apply.py
@@ -79,6 +79,15 @@ def test_no_auto_apply_low_confidence(self):
         result = self.move._ai_can_auto_apply(data)
         self.assertFalse(result)
 
+    def test_no_auto_apply_when_preview_review_required(self):
+        """Preview-only results must not bypass the manual review flow."""
+        data = self._make_high_confidence_data()
+        data['_source'] = 'facturx'
+        data['_require_preview'] = True
+        result = self.move._ai_can_auto_apply(data)
+        self.assertFalse(result)
+        self.assertNotIn('_force_partner_id', data)
+
     def test_no_auto_apply_unknown_vendor(self):
         """Unknown vendor (no VAT match) → auto-apply returns False."""
         data = self._make_high_confidence_data()
diff --git a/tests/test_facturx_pipeline.py b/tests/test_facturx_pipeline.py
@@ -1,20 +1,60 @@
-"""Factur-X pipeline integration tests.
-
-Covers the Factur-X shortcut in the extraction pipeline:
-  1. Factur-X detection skips AI API call
-  2. Returns correct wrapper dict
-  3. Non-PDF and unavailable cases
-  4. Apply and preview modes
-"""
+"""Factur-X pipeline integration tests."""
 
 import base64
+import json
 from unittest.mock import MagicMock, patch
 
 from odoo.tests.common import TransactionCase, tagged
 
 _MODULE = 'odoo.addons.account_invoice_digitize_ai'
 
-SAMPLE_FACTURX_XML = '<xml>facturx</xml>'
+SAMPLE_FACTURX_XML = b'<xml>facturx</xml>'
+SAMPLE_FACTURX_DATA = {
+    'document_type': 'invoice',
+    'is_marked_paid': False,
+    'vendor': {'name': 'Factur-X Vendor', 'vat': 'DE123456789', 'confidence': 1.0},
+    'buyer': {'name': 'Factor 3', 'vat': 'DE999999999', 'confidence': 1.0},
+    'invoice': {
+        'reference': 'FX-001',
+        'invoice_date': '2024-01-15',
+        'currency': 'EUR',
+        'confidence': 1.0,
+    },
+    'totals': {
+        'untaxed_amount': 1000.0,
+        'tax_amount': 200.0,
+        'total_amount': 1200.0,
+        'confidence': 1.0,
+    },
+    'tax_lines': [
+        {
+            'tax_label': 'VAT 20%',
+            'tax_rate': 20.0,
+            'base_amount': 1000.0,
+            'tax_amount': 200.0,
+            'confidence': 1.0,
+        }
+    ],
+    'lines': [
+        {
+            'description': 'Consulting services',
+            'quantity': 1.0,
+            'unit_price': 1000.0,
+            'subtotal_untaxed': 1000.0,
+            'tax_rate': 20.0,
+        }
+    ],
+    'table_analysis': {
+        'number_format': 'dot_decimal',
+        'complexity': 'simple',
+        'line_count': 1,
+    },
+}
+SAMPLE_FACTURX_PREVIEW_DATA = {
+    **SAMPLE_FACTURX_DATA,
+    '_source': 'facturx',
+    '_require_preview': True,
+}
 
 
 @tagged('post_install', '-at_install')
@@ -91,39 +131,116 @@ def test_facturx_unavailable_skipped(self):
     @patch(f'{_MODULE}.models.ai_facturx_parser.parse_facturx_xml')
     def test_facturx_apply_sets_done(self, mock_parse, _is_pdf, _detect):
         """Successful Factur-X extraction should set status to 'done'."""
-        mock_parse.return_value = {
-            'vendor': {'name': 'Test Vendor', 'confidence': 1.0},
-            'invoice': {'number': 'FX-001', 'date': '2024-01-15', 'confidence': 1.0},
-            'totals': {'total_amount': 1200, 'confidence': 1.0},
-            'document_type': 'invoice',
-        }
+        mock_parse.return_value = SAMPLE_FACTURX_DATA
+        move = self.env['account.move'].create({
+            'move_type': 'in_invoice',
+            'company_id': self.company.id,
+        })
         attachment = self.env['ir.attachment'].create({
             'name': 'test.pdf',
             'datas': base64.b64encode(b'fake-pdf'),
             'res_model': 'account.move',
-            'res_id': self.move.id,
+            'res_id': move.id,
         })
-        self.move._ai_trigger_extraction('test-key-123', attachment)
-        self.assertEqual(self.move.ai_extraction_status, 'done')
+        move._ai_trigger_extraction('test-key-123', attachment)
+        self.assertEqual(move.ai_extraction_status, 'done')
 
-    # --- 6. Preview returns data without applying ---
+    # --- 6. Preview returns normalized data without applying ---
 
     @patch(f'{_MODULE}.models.ai_document.detect_facturx', return_value=SAMPLE_FACTURX_XML)
     @patch(f'{_MODULE}.models.ai_document.FACTURX_AVAILABLE', True)
     @patch(f'{_MODULE}.models.ai_document.is_pdf', return_value=True)
-    def test_facturx_preview_returns_data(self, _is_pdf, _detect):
-        """In preview mode, Factur-X data should be returned without applying."""
+    @patch(f'{_MODULE}.models.ai_facturx_parser.parse_facturx_xml', return_value=SAMPLE_FACTURX_DATA)
+    def test_facturx_preview_returns_normalized_data(self, mock_parse, _is_pdf, _detect):
+        """Preview mode should normalize Factur-X XML into JSON-safe extraction data."""
+        move = self.env['account.move'].create({
+            'move_type': 'in_invoice',
+            'company_id': self.company.id,
+        })
         attachment = self.env['ir.attachment'].create({
             'name': 'test.pdf',
             'datas': base64.b64encode(b'fake-pdf'),
             'res_model': 'account.move',
-            'res_id': self.move.id,
+            'res_id': move.id,
         })
-        result = self.move._ai_trigger_extraction(
+        result = move._ai_trigger_extraction(
             'test-key-123', attachment, preview=True,
         )
+        mock_parse.assert_called_once_with(SAMPLE_FACTURX_XML)
         self.assertIsNotNone(result)
-        self.assertTrue(result.get('_facturx'))
-        self.assertEqual(result['_xml'], SAMPLE_FACTURX_XML)
+        self.assertEqual(result, SAMPLE_FACTURX_PREVIEW_DATA)
         # Status should NOT be 'done' (preview mode)
-        self.assertNotEqual(self.move.ai_extraction_status, 'done')
+        self.assertNotEqual(move.ai_extraction_status, 'done')
+
+    # --- 7. Sync button uses normalized Factur-X preview data ---
+
+    @patch(f'{_MODULE}.models.ai_document.detect_facturx', return_value=SAMPLE_FACTURX_XML)
+    @patch(f'{_MODULE}.models.ai_document.FACTURX_AVAILABLE', True)
+    @patch(f'{_MODULE}.models.ai_document.is_pdf', return_value=True)
+    @patch(f'{_MODULE}.models.ai_facturx_parser.parse_facturx_xml', return_value=SAMPLE_FACTURX_DATA)
+    def test_facturx_sync_button_opens_preview_wizard(self, mock_parse, _is_pdf, _detect):
+        """The normal button path must open the preview wizard for Factur-X PDFs."""
+        move = self.env['account.move'].create({
+            'move_type': 'in_invoice',
+            'company_id': self.company.id,
+        })
+        self.env['ir.attachment'].create({
+            'name': 'test.pdf',
+            'datas': base64.b64encode(b'fake-pdf'),
+            'mimetype': 'application/pdf',
+            'res_model': 'account.move',
+            'res_id': move.id,
+        })
+
+        result = move.action_ai_extract()
+
+        mock_parse.assert_called_once_with(SAMPLE_FACTURX_XML)
+        self.assertEqual(result.get('res_model'), 'ai.preview.wizard')
+        self.assertEqual(move.ai_extraction_status, 'pending')
+        self.assertEqual(json.loads(move.ai_last_extraction_data), SAMPLE_FACTURX_PREVIEW_DATA)
+
+        wizard = self.env['ai.preview.wizard'].browse(result['res_id'])
+        self.assertEqual(wizard.invoice_ref, 'FX-001')
+        self.assertEqual(wizard.vendor_name, 'Factur-X Vendor')
+
+    @patch(f'{_MODULE}.models.ai_document.detect_facturx', return_value=SAMPLE_FACTURX_XML)
+    @patch(f'{_MODULE}.models.ai_document.FACTURX_AVAILABLE', True)
+    @patch(f'{_MODULE}.models.ai_document.is_pdf', return_value=True)
+    @patch(f'{_MODULE}.models.ai_facturx_parser.parse_facturx_xml', return_value=SAMPLE_FACTURX_DATA)
+    def test_facturx_sync_button_skips_auto_apply_and_keeps_preview(self, mock_parse, _is_pdf, _detect):
+        """Factur-X preview data must not be auto-applied even for reliable vendors."""
+        self.env['ir.config_parameter'].sudo().set_param(
+            'account_invoice_digitize_ai.ai_auto_apply_enabled',
+            'True',
+        )
+        partner = self.env['res.partner'].create({
+            'name': 'Factur-X Vendor',
+            'is_company': True,
+            'vat': 'DE123456789',
+        })
+        self.env['ai.vendor.score'].create({
+            'partner_id': partner.id,
+            'company_id': self.company.id,
+            'total_extractions': 10,
+            'correct_extractions': 9,
+        })
+        move = self.env['account.move'].create({
+            'move_type': 'in_invoice',
+            'company_id': self.company.id,
+        })
+        self.env['ir.attachment'].create({
+            'name': 'test.pdf',
+            'datas': base64.b64encode(b'fake-pdf'),
+            'mimetype': 'application/pdf',
+            'res_model': 'account.move',
+            'res_id': move.id,
+        })
+
+        result = move.action_ai_extract()
+
+        mock_parse.assert_called_once_with(SAMPLE_FACTURX_XML)
+        self.assertEqual(result.get('res_model'), 'ai.preview.wizard')
+        self.assertEqual(move.ai_extraction_status, 'pending')
+        self.assertFalse(move.ref)
+        self.assertFalse(move.partner_id)
+        self.assertEqual(json.loads(move.ai_last_extraction_data), SAMPLE_FACTURX_PREVIEW_DATA)
diff --git a/tests/test_preprocessing.py b/tests/test_preprocessing.py
@@ -543,11 +543,11 @@ def test_facturx_takes_priority(self):
         ICP.set_param('account_invoice_digitize_ai.ai_azure_api_key', 'test-azure-key')
 
         move = self._create_move()
-        facturx_data = {'vendor': {}, 'invoice': {}, 'totals': {}}
+        facturx_xml = b'<xml>facturx</xml>'
 
         AccountMove = type(move)
         with (
-            patch.object(AccountMove, '_ai_try_facturx', return_value=facturx_data),
+            patch.object(AccountMove, '_ai_try_facturx', return_value=facturx_xml),
             patch.object(AccountMove, '_ai_apply_facturx') as mock_fx,
             patch.object(AccountMove, '_ai_try_preprocess') as mock_pp,
         ):
