Workaround to fix missing db fields

cope2n-api/fwd_api/api/ctel_view.py

@@ -214,14 +214,6 @@ class CtelViewSet(viewsets.ViewSet):
             "imei": imei_file_objs,
             "invoice": invoice_file_objs
         }
-        total_page = len(files.keys())
-        p_type = validated_data['type']
-        new_request: SubscriptionRequest = SubscriptionRequest(pages=total_page,
-                                                            pages_left=total_page,
-                                                            process_type=p_type, status=1, request_id=rq_id,
-                                                            provider_code=provider_code,
-                                                            subscription=sub)
-        new_request.save()
 
         count = 0
         doc_files_with_type = []
@@ -236,6 +228,15 @@ class CtelViewSet(viewsets.ViewSet):
                 ))
                 count += 1
 
+        total_page = len(doc_files_with_type)
+        p_type = validated_data['type']
+        new_request: SubscriptionRequest = SubscriptionRequest(pages=total_page,
+                                                            pages_left=total_page,
+                                                            process_type=p_type, status=1, request_id=rq_id,
+                                                            provider_code=provider_code,
+                                                            subscription=sub)
+        new_request.save()
+
         # Run file processing in a pool of 2 threads. TODO: Convert to Celery worker when possible
         compact_files = [None] * len(doc_files_with_type)
         pool = ThreadPool(processes=2)

cope2n-api/fwd_api/celery_worker/internal_task.py

@@ -13,6 +13,8 @@ from fwd_api.models import SubscriptionRequestFile
 from ..utils import file as FileUtils
 from ..utils import process as ProcessUtil
 from ..utils import s3 as S3Util
+from fwd_api.constant.common import ProcessType
+
 
 from celery.utils.log import get_task_logger
 from fwd import settings
@@ -113,6 +115,24 @@ def process_pdf(rq_id, sub_id, p_type, user_id, files):
     new_request.ai_inference_start_time = time.time()
     new_request.save()
 
+    trials = 0
+    while True:
+        rq: SubscriptionRequest = \
+            SubscriptionRequest.objects.filter(request_id=rq_id).first()
+        if rq.ai_inference_start_time != 0:
+            break
+        time.sleep(0.1)
+        trials += 1
+        if trials > 5:
+            rq.preprocessing_time = time.time() - start_time
+            rq.doc_type = doc_type_string
+            rq.ai_inference_start_time = time.time()
+            rq.save()
+        if trials > 10:
+            rq.status = 404
+            rq.save()
+            return
+
     # Send to next queue
     for sub_rq_id, sub_id, urls, user_id, p_type in to_queue:
         ProcessUtil.send_to_queue2(sub_rq_id, sub_id, urls, user_id, p_type)

cope2n-api/fwd_api/celery_worker/process_result_tasks.py

@@ -145,8 +145,7 @@ def process_invoice_sbt_result(rq_id, result):
     try:
         page_index = int(rq_id.split("_sub_")[1])
         rq_id = rq_id.split("_sub_")[0]
-        rq: SubscriptionRequest = \
+        rq: SubscriptionRequest = SubscriptionRequest.objects.filter(request_id=rq_id).first()
-            SubscriptionRequest.objects.filter(request_id=rq_id, process_type=ProcessType.SBT_INVOICE.value)[0]
         for i in range(10):
             if rq.ai_inference_start_time == 0:
                 logging.warn(f"ai_inference_start_time = 0, looks like database is lagging, attemp {i} in 0.2 second ...")