feat: add DocumentTreeReadAPI for document pagination and file size limit to knowledge model

liuruibin · liuruibin · commit 6f5645e16c24 · 2025-05-06T17:39:26.000+08:00
diff --git a/apps/knowledge/api/document.py b/apps/knowledge/api/document.py
@@ -266,3 +266,52 @@ def get_parameters():
     @staticmethod
     def get_request():
         return BatchEditHitHandlingSerializer
+
+
+class DocumentTreeReadAPI(APIMixin):
+    @staticmethod
+    def get_parameters():
+        return [
+            OpenApiParameter(
+                name="workspace_id",
+                description="工作空间id",
+                type=OpenApiTypes.STR,
+                location='path',
+                required=True,
+            ),
+            OpenApiParameter(
+                name="knowledge_id",
+                description="知识库id",
+                type=OpenApiTypes.STR,
+                location='path',
+                required=True,
+            ),
+            OpenApiParameter(
+                name="folder_id",
+                description="文件夹id",
+                type=OpenApiTypes.STR,
+                location='query',
+                required=True,
+            ),
+            OpenApiParameter(
+                name="user_id",
+                description="用户id",
+                type=OpenApiTypes.STR,
+                location='query',
+                required=False,
+            ),
+            OpenApiParameter(
+                name="name",
+                description="名称",
+                type=OpenApiTypes.STR,
+                location='query',
+                required=False,
+            ),
+            OpenApiParameter(
+                name="desc",
+                description="描述",
+                type=OpenApiTypes.STR,
+                location='query',
+                required=False,
+            ),
+        ]
diff --git a/apps/knowledge/migrations/0004_knowledge_file_size_limit_alter_document_status_and_more.py b/apps/knowledge/migrations/0004_knowledge_file_size_limit_alter_document_status_and_more.py
@@ -0,0 +1,29 @@
+# Generated by Django 5.2 on 2025-05-06 09:05
+
+import knowledge.models.knowledge
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('knowledge', '0003_alter_document_status_alter_paragraph_status_and_more'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='knowledge',
+            name='file_size_limit',
+            field=models.IntegerField(default=100, verbose_name='文件大小限制'),
+        ),
+        migrations.AlterField(
+            model_name='document',
+            name='status',
+            field=models.CharField(default=knowledge.models.knowledge.Status.__str__, max_length=20, verbose_name='状态'),
+        ),
+        migrations.AlterField(
+            model_name='paragraph',
+            name='status',
+            field=models.CharField(default=knowledge.models.knowledge.Status.__str__, max_length=20, verbose_name='状态'),
+        ),
+    ]
diff --git a/apps/knowledge/models/knowledge.py b/apps/knowledge/models/knowledge.py
@@ -127,6 +127,7 @@ class Knowledge(AppModelMixin):
     folder = models.ForeignKey(KnowledgeFolder, on_delete=models.DO_NOTHING, verbose_name="文件夹id", default='root')
     embedding_model = models.ForeignKey(Model, on_delete=models.DO_NOTHING, verbose_name="向量模型",
                                         default=default_model)
+    file_size_limit = models.IntegerField(verbose_name="文件大小限制", default=100)
     meta = models.JSONField(verbose_name="元数据", default=dict)
 
     class Meta:
@@ -197,6 +198,7 @@ class ProblemParagraphMapping(AppModelMixin):
     class Meta:
         db_table = "problem_paragraph_mapping"
 
+
 class ApplicationKnowledgeMapping(AppModelMixin):
     id = models.UUIDField(primary_key=True, max_length=128, default=uuid.uuid7, editable=False, verbose_name="主键id")
     # application = models.ForeignKey(Application, on_delete=models.CASCADE)
@@ -206,7 +208,6 @@ class Meta:
         db_table = "application_knowledge_mapping"
 
 
-
 class SourceType(models.IntegerChoices):
     """订单类型"""
     PROBLEM = 0, '问题'
diff --git a/apps/knowledge/urls.py b/apps/knowledge/urls.py
@@ -20,5 +20,6 @@
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/refresh', views.DocumentView.Refresh.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/cancel_task', views.DocumentView.CancelTask.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/cancel_task/batch', views.DocumentView.BatchCancelTask.as_view()),
+    path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<int:current_page>/<int:page_sige>', views.DocumentView.Page.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<int:current_page>/<int:page_size>', views.KnowledgeView.Page.as_view()),
 ]
diff --git a/apps/knowledge/views/document.py b/apps/knowledge/views/document.py
@@ -10,8 +10,8 @@
 from common.result import result
 from knowledge.api.document import DocumentSplitAPI, DocumentBatchAPI, DocumentBatchCreateAPI, DocumentCreateAPI, \
     DocumentReadAPI, DocumentEditAPI, DocumentDeleteAPI, TableDocumentCreateAPI, QaDocumentCreateAPI, \
-    WebDocumentCreateAPI, CancelTaskAPI, BatchCancelTaskAPI, SyncWebAPI, RefreshAPI, BatchEditHitHandlingAPI
-from knowledge.api.knowledge import KnowledgeTreeReadAPI
+    WebDocumentCreateAPI, CancelTaskAPI, BatchCancelTaskAPI, SyncWebAPI, RefreshAPI, BatchEditHitHandlingAPI, \
+    DocumentTreeReadAPI
 from knowledge.serializers.document import DocumentSerializers
 
 
@@ -40,8 +40,8 @@ def post(self, request: Request, workspace_id: str, knowledge_id: str):
         description=_('Get document'),
         summary=_('Get document'),
         operation_id=_('Get document'),
-        parameters=KnowledgeTreeReadAPI.get_parameters(),
-        responses=KnowledgeTreeReadAPI.get_response(),
+        parameters=DocumentTreeReadAPI.get_parameters(),
+        responses=DocumentTreeReadAPI.get_response(),
         tags=[_('Knowledge Base/Documentation')]
     )
     @has_permissions(PermissionConstants.DOCUMENT_READ.get_workspace_permission())
@@ -163,16 +163,17 @@ class SyncWeb(APIView):
         authentication_classes = [TokenAuth]
 
         @extend_schema(
-            methods=['PUT'],
+            methods=['GET'],
             description=_('Synchronize web site types'),
             summary=_('Synchronize web site types'),
             operation_id=_('Synchronize web site types'),
             parameters=SyncWebAPI.get_parameters(),
+            request=SyncWebAPI.get_request(),
             responses=SyncWebAPI.get_response(),
             tags=[_('Knowledge Base/Documentation')]
         )
         @has_permissions(PermissionConstants.DOCUMENT_EDIT.get_workspace_permission())
-        def put(self, request: Request, workspace_id: str, knowledge_id: str, document_id: str):
+        def get(self, request: Request, workspace_id: str, knowledge_id: str, document_id: str):
             return result.success(DocumentSerializers.Sync(
                 data={'document_id': document_id, 'knowledge_id': knowledge_id, 'workspace_id': workspace_id}
             ).sync())
@@ -291,6 +292,31 @@ def delete(self, request: Request, workspace_id: str, knowledge_id: str):
                 data={'workspace_id': workspace_id, 'knowledge_id': knowledge_id}
             ).batch_delete(request.data))
 
+    class Page(APIView):
+        authentication_classes = [TokenAuth]
+
+        @extend_schema(
+            methods=['GET'],
+            description=_('Get document by pagination'),
+            summary=_('Get document by pagination'),
+            operation_id=_('Get document by pagination'),
+            parameters=DocumentTreeReadAPI.get_parameters(),
+            responses=DocumentTreeReadAPI.get_response(),
+            tags=[_('Knowledge Base/Documentation')]
+        )
+        @has_permissions(PermissionConstants.DOCUMENT_READ.get_workspace_permission())
+        def get(self, request: Request, workspace_id: str, knowledge_id: str, current_page: int, page_size: int):
+            return result.success(DocumentSerializers.Query(
+                data={
+                    'workspace_id': workspace_id,
+                    'knowledge_id': knowledge_id,
+                    'folder_id': request.query_params.get('folder_id'),
+                    'name': request.query_params.get('name'),
+                    'desc': request.query_params.get("desc"),
+                    'user_id': request.query_params.get('user_id')
+                }
+            ).page(current_page, page_size))
+
 
 class WebDocumentView(APIView):
     authentication_classes = [TokenAuth]

Original file line number	Diff line number	Diff line change
`@@ -20,5 +20,6 @@`
`20`	`20`	`path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/refresh', views.DocumentView.Refresh.as_view()),`
`21`	`21`	`path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/cancel_task', views.DocumentView.CancelTask.as_view()),`
`22`	`22`	`path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<str:document_id>/cancel_task/batch', views.DocumentView.BatchCancelTask.as_view()),`
	`23`	`+ path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/<int:current_page>/<int:page_sige>', views.DocumentView.Page.as_view()),`
`23`	`24`	`path('workspace/<str:workspace_id>/knowledge/<int:current_page>/<int:page_size>', views.KnowledgeView.Page.as_view()),`
`24`	`25`	`]`