fix: 修复批量删除文档没有删除问题的缺陷

--bug=1048687 --user=王孝刚 【知识库】删除文档不会删除文档中分段关联的问题 https://www.tapd.cn/57709429/s/1624544

Author: wxg0103

apps/dataset/serializers/document_serializers.py

@@ -1091,17 +1091,15 @@ def file_to_paragraph(file, pattern_list: List, with_filter: bool, limit: int):
 
 
 def delete_problems_and_mappings(document_ids):
-    # 获取所有需要删除的问题ID
-    problem_ids = list(
-        ProblemParagraphMapping.objects.filter(document_id__in=document_ids).values_list('problem_id', flat=True))
+    problem_paragraph_mappings = ProblemParagraphMapping.objects.filter(document_id__in=document_ids)
+    problem_ids = set(problem_paragraph_mappings.values_list('problem_id', flat=True))
 
     if problem_ids:
-        ProblemParagraphMapping.objects.filter(document_id__in=document_ids).delete()
+        problem_paragraph_mappings.delete()
         remaining_problem_counts = ProblemParagraphMapping.objects.filter(problem_id__in=problem_ids).values(
             'problem_id').annotate(count=Count('problem_id'))
-
-        problem_ids_to_delete = [pid for pid in problem_ids if
-                                 not any(pc['problem_id'] == pid for pc in remaining_problem_counts)]
+        remaining_problem_ids = {pc['problem_id'] for pc in remaining_problem_counts}
+        problem_ids_to_delete = problem_ids - remaining_problem_ids
         Problem.objects.filter(id__in=problem_ids_to_delete).delete()
     else:
-        ProblemParagraphMapping.objects.filter(document_id__in=document_ids).delete()
+        problem_paragraph_mappings.delete()

apps/dataset/serializers/paragraph_serializers.py

@@ -751,16 +751,15 @@ def batch_generate_related(self, instance: Dict, with_valid=True):
 
 
 def delete_problems_and_mappings(paragraph_ids):
-    problem_ids = list(
-        ProblemParagraphMapping.objects.filter(paragraph_id__in=paragraph_ids).values_list('problem_id', flat=True))
+    problem_paragraph_mappings = ProblemParagraphMapping.objects.filter(paragraph_id__in=paragraph_ids)
+    problem_ids = set(problem_paragraph_mappings.values_list('problem_id', flat=True))
 
     if problem_ids:
-        ProblemParagraphMapping.objects.filter(paragraph_id__in=paragraph_ids).delete()
+        problem_paragraph_mappings.delete()
         remaining_problem_counts = ProblemParagraphMapping.objects.filter(problem_id__in=problem_ids).values(
             'problem_id').annotate(count=Count('problem_id'))
-
-        problem_ids_to_delete = [pid for pid in problem_ids if
-                                 not any(pc['problem_id'] == pid for pc in remaining_problem_counts)]
+        remaining_problem_ids = {pc['problem_id'] for pc in remaining_problem_counts}
+        problem_ids_to_delete = problem_ids - remaining_problem_ids
         Problem.objects.filter(id__in=problem_ids_to_delete).delete()
     else:
-        ProblemParagraphMapping.objects.filter(paragraph_id__in=paragraph_ids).delete()
+        problem_paragraph_mappings.delete()