fix: When deleting conversation logs, they should be deleted by conversation record, not by session.

--bug=1051378 --user=王孝刚 【应用】对话日志删除时应该按对话记录删除，不能按会话删除 https://www.tapd.cn/57709429/s/1645443

Author: wxg0103

apps/common/job/clean_chat_job.py

@@ -7,12 +7,11 @@
 from django.utils import timezone
 from apscheduler.schedulers.background import BackgroundScheduler
 from django_apscheduler.jobstores import DjangoJobStore
-from application.models import Application, Chat
-from django.db.models import Q
+from application.models import Application, Chat, ChatRecord
+from django.db.models import Q, Max
 from common.lock.impl.file_lock import FileLock
 from dataset.models import File
-
-
+from django.db import connection
 
 scheduler = BackgroundScheduler()
 scheduler.add_jobstore(DjangoJobStore(), "default")
@@ -32,23 +31,42 @@ def clean_chat_log_job():
 
     query_conditions = Q()
     for app_id, cutoff_date in cutoff_dates.items():
-        query_conditions |= Q(application_id=app_id, create_time__lt=cutoff_date)
-
+        query_conditions |= Q(chat__application_id=app_id, create_time__lt=cutoff_date)
     batch_size = 500
     while True:
         with transaction.atomic():
-            logs_to_delete = Chat.objects.filter(query_conditions).values_list('id', flat=True)[:batch_size]
-            count = logs_to_delete.count()
-            logs_to_delete_str = [str(uuid) for uuid in logs_to_delete]
-            if count == 0:
+            chat_records = ChatRecord.objects.filter(query_conditions).select_related('chat').only('id', 'chat_id',
+                                                                                                   'create_time')[
+                           :batch_size]
+            if not chat_records:
                 break
-            deleted_count, _ = Chat.objects.filter(id__in=logs_to_delete).delete()
-            # 删除对应的文件
-            File.objects.filter(meta__chat_id__in=logs_to_delete_str).delete()
+            chat_record_ids = [record.id for record in chat_records]
+            chat_ids = {record.chat_id for record in chat_records}
+
+            # 计算每个 chat_id 的最大 create_time
+            max_create_times = ChatRecord.objects.filter(id__in=chat_record_ids).values('chat_id').annotate(
+                max_create_time=Max('create_time'))
+
+            # 收集需要删除的文件
+            files_to_delete = []
+            for record in chat_records:
+                max_create_time = next(
+                    (item['max_create_time'] for item in max_create_times if item['chat_id'] == record.chat_id), None)
+                if max_create_time:
+                    files_to_delete.extend(
+                        File.objects.filter(meta__chat_id=str(record.chat_id), create_time__lt=max_create_time)
+                    )
+            # 删除 ChatRecord
+            deleted_count = ChatRecord.objects.filter(id__in=chat_record_ids).delete()[0]
+
+            # 删除没有关联 ChatRecord 的 Chat
+            Chat.objects.filter(chatrecord__isnull=True, id__in=chat_ids).delete()
+            File.objects.filter(loid__in=[file.loid for file in files_to_delete]).delete()
+
             if deleted_count < batch_size:
                 break
 
-    logging.getLogger("max_kb").info(_('end clean chat log'))
+    logging.getLogger("max_kb").info(f'结束清理对话记录')
 
 
 def run():