fix: keep queued async extraction bound to one attachment

Author: Florian Wunderlich

models/account_move.py

@@ -90,6 +90,13 @@ class AccountMove(models.Model):
         ondelete='set null',
         help='Attachment used for the cached extraction.',
     )
+    ai_queued_attachment_id = fields.Many2one(
+        'ir.attachment',
+        string='Queued Attachment',
+        copy=False,
+        ondelete='set null',
+        help='Attachment selected for the next background extraction run.',
+    )
     ai_extraction_queued_at = fields.Datetime(
         string='Queued At',
         copy=False,
@@ -386,6 +393,7 @@ def action_ai_extract(self):
 
         # --- Async mode: queue for background processing -------------------
         if self._ai_get_bool_param('ai_async_extraction'):
+            self.ai_queued_attachment_id = attachment.id
             self.ai_extraction_status = 'processing'
             self.ai_extraction_queued_at = fields.Datetime.now()
             return self._ai_notify(

models/ai_cron_processor.py

@@ -29,7 +29,11 @@ def _ai_cron_process_queue(self):
         ])
         if stale:
             _logger.warning('AI cron: marking %d stale extractions as failed', len(stale))
-            stale.write({'ai_extraction_status': 'failed', 'ai_extraction_queued_at': False})
+            stale.write({
+                'ai_extraction_status': 'failed',
+                'ai_extraction_queued_at': False,
+                'ai_queued_attachment_id': False,
+            })
 
         # Claim the batch with a row-level lock so a concurrent transaction
         # (a manual extraction, or a second trigger) cannot grab the same
@@ -65,15 +69,21 @@ def _ai_cron_process_queue(self):
                 _logger.exception('AI cron: extraction failed for move %s', move.id)
                 move.ai_extraction_status = 'failed'
                 move.ai_extraction_queued_at = False
+                move.ai_queued_attachment_id = False
             # Commit after each invoice to avoid losing work on error
             self.env.cr.commit()  # noqa: B010
 
     def _ai_cron_extract_one(self, move, api_key):
         """Process a single queued extraction."""
-        attachment = move._ai_get_invoice_attachment()
+        attachment = (
+            move.ai_queued_attachment_id
+            or move.ai_last_extraction_attachment_id
+            or move._ai_get_invoice_attachment()
+        )
         if not attachment:
             move.ai_extraction_status = 'failed'
             move.ai_extraction_queued_at = False
+            move.ai_queued_attachment_id = False
             return
 
         data = move._ai_trigger_extraction(api_key, attachment, preview=True)
@@ -96,3 +106,4 @@ def _ai_cron_extract_one(self, move, api_key):
                     )
             move.ai_extraction_status = 'done'
         move.ai_extraction_queued_at = False
+        move.ai_queued_attachment_id = False

tests/test_async_extraction.py

@@ -62,11 +62,14 @@ def test_button_queues_extraction(self):
         """Button should set status='processing' + queued_at when async enabled."""
         self.ICP.set_param(self._p + 'ai_async_extraction', 'True')
         move = self._create_invoice_with_attachment()
+        attachment = move.attachment_ids.filtered(lambda att: att.mimetype == 'application/pdf')[:1]
 
         result = move.action_ai_extract()
 
         self.assertEqual(move.ai_extraction_status, 'processing')
         self.assertTrue(move.ai_extraction_queued_at)
+        self.assertEqual(move.ai_queued_attachment_id, attachment)
+        self.assertFalse(move.ai_last_extraction_attachment_id)
         self.assertEqual(result['tag'], 'display_notification')
 
     def test_cron_processes_queue(self):
@@ -85,6 +88,39 @@ def test_cron_processes_queue(self):
         self.assertEqual(move.ai_extraction_status, 'done')
         self.assertTrue(move.ai_last_extraction_data)
         self.assertFalse(move.ai_extraction_queued_at)
+        self.assertFalse(move.ai_queued_attachment_id)
+
+    def test_cron_uses_scheduled_attachment(self):
+        """Cron should process the attachment that was queued, not rediscover a newer one."""
+        self.ICP.set_param(self._p + 'ai_async_extraction', 'True')
+        move = self._create_invoice_with_attachment()
+        queued_attachment = move.attachment_ids.filtered(lambda att: att.mimetype == 'application/pdf')[:1]
+        self.env['ir.attachment'].create(
+            {
+                'name': 'later.png',
+                'datas': base64.b64encode(b'fake-image'),
+                'res_model': 'account.move',
+                'res_id': move.id,
+                'mimetype': 'image/png',
+            }
+        )
+        move.ai_extraction_status = 'processing'
+        move.ai_extraction_queued_at = fields.Datetime.now()
+        move.ai_queued_attachment_id = queued_attachment.id
+
+        def _mock_extract(api_key, attachment, preview=False):
+            self.assertEqual(attachment, queued_attachment)
+            return self.mock_data
+
+        with patch.object(
+            type(move),
+            '_ai_trigger_extraction',
+            side_effect=_mock_extract,
+        ), patch.object(self.env.cr, 'commit'):
+            self.env['account.move']._ai_cron_process_queue()
+
+        self.assertEqual(move.ai_last_extraction_attachment_id, queued_attachment)
+        self.assertFalse(move.ai_queued_attachment_id)
 
     def test_cron_handles_failure(self):
         """Cron should set status='failed' when extraction returns None."""
@@ -101,6 +137,7 @@ def test_cron_handles_failure(self):
 
         self.assertEqual(move.ai_extraction_status, 'failed')
         self.assertFalse(move.ai_extraction_queued_at)
+        self.assertFalse(move.ai_queued_attachment_id)
 
     def test_sync_fallback(self):
         """When async is disabled, extraction should happen synchronously."""
@@ -144,14 +181,17 @@ def test_cron_marks_stale_as_failed(self):
         """Items queued > 10 minutes ago should be marked as failed."""
         self.ICP.set_param(self._p + 'ai_async_extraction', 'True')
         move = self._create_invoice_with_attachment()
+        queued_attachment = move.attachment_ids.filtered(lambda att: att.mimetype == 'application/pdf')[:1]
         move.ai_extraction_status = 'processing'
         move.ai_extraction_queued_at = fields.Datetime.now() - timedelta(minutes=15)
+        move.ai_queued_attachment_id = queued_attachment.id
 
         with patch.object(self.env.cr, 'commit'):
             self.env['account.move']._ai_cron_process_queue()
 
         self.assertEqual(move.ai_extraction_status, 'failed')
         self.assertFalse(move.ai_extraction_queued_at)
+        self.assertFalse(move.ai_queued_attachment_id)
 
     def test_cron_batch_size_limit(self):
         """Cron should only process up to 5 items per run."""