feat: add DocumentSplitPatternAPI for segment ID retrieval with workspace and knowledge ID parameters

Author: liuruibin

apps/knowledge/api/document.py

@@ -315,3 +315,28 @@ def get_parameters():
                 required=False,
             ),
         ]
+
+
+class DocumentSplitPatternAPI(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,
+            ),
+        ]
+
+    @staticmethod
+    def get_response():
+        return DefaultResultSerializer

apps/knowledge/serializers/document.py

@@ -624,6 +624,26 @@ def file_to_paragraph(self, file, pattern_list: List, with_filter: bool, limit:
                 return result
             return [result]
 
+    class SplitPattern(serializers.Serializer):
+        workspace_id = serializers.CharField(required=True, label=_('workspace id'))
+        knowledge_id = serializers.UUIDField(required=True, label=_('knowledge id'))
+
+        @staticmethod
+        def list():
+            return [
+                {'key': "#", 'value': '(?<=^)# .*|(?<=\\n)# .*'},
+                {'key': '##', 'value': '(?<=\\n)(?<!#)## (?!#).*|(?<=^)(?<!#)## (?!#).*'},
+                {'key': '###', 'value': "(?<=\\n)(?<!#)### (?!#).*|(?<=^)(?<!#)### (?!#).*"},
+                {'key': '####', 'value': "(?<=\\n)(?<!#)#### (?!#).*|(?<=^)(?<!#)#### (?!#).*"},
+                {'key': '#####', 'value': "(?<=\\n)(?<!#)##### (?!#).*|(?<=^)(?<!#)##### (?!#).*"},
+                {'key': '######', 'value': "(?<=\\n)(?<!#)###### (?!#).*|(?<=^)(?<!#)###### (?!#).*"},
+                {'key': '-', 'value': '(?<! )- .*'},
+                {'key': _('space'), 'value': '(?<! ) (?! )'},
+                {'key': _('semicolon'), 'value': '(?<!；)；(?!；)'}, {'key': _('comma'), 'value': '(?<!，)，(?!，)'},
+                {'key': _('period'), 'value': '(?<!。)。(?!。)'}, {'key': _('enter'), 'value': '(?<!\\n)\\n(?!\\n)'},
+                {'key': _('blank line'), 'value': '(?<!\\n)\\n\\n(?!\\n)'}
+            ]
+
     class Batch(serializers.Serializer):
         workspace_id = serializers.UUIDField(required=True, label=_('workspace id'))
         knowledge_id = serializers.UUIDField(required=True, label=_('knowledge id'))

apps/knowledge/urls.py

@@ -10,6 +10,7 @@
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>', views.KnowledgeView.Operate.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document', views.DocumentView.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/split', views.DocumentView.Split.as_view()),
+    path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/split_pattern', views.DocumentView.SplitPattern.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/batch', views.DocumentView.Batch.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/web', views.WebDocumentView.as_view()),
     path('workspace/<str:workspace_id>/knowledge/<str:knowledge_id>/document/qa', views.QaDocumentView.as_view()),

apps/knowledge/views/document.py

@@ -11,7 +11,7 @@
 from knowledge.api.document import DocumentSplitAPI, DocumentBatchAPI, DocumentBatchCreateAPI, DocumentCreateAPI, \
     DocumentReadAPI, DocumentEditAPI, DocumentDeleteAPI, TableDocumentCreateAPI, QaDocumentCreateAPI, \
     WebDocumentCreateAPI, CancelTaskAPI, BatchCancelTaskAPI, SyncWebAPI, RefreshAPI, BatchEditHitHandlingAPI, \
-    DocumentTreeReadAPI
+    DocumentTreeReadAPI, DocumentSplitPatternAPI
 from knowledge.serializers.document import DocumentSerializers
 
 
@@ -140,6 +140,22 @@ def post(self, request: Request, workspace_id: str, knowledge_id: str):
                 'knowledge_id': knowledge_id,
             }).parse(split_data))
 
+    class SplitPattern(APIView):
+        authentication_classes = [TokenAuth]
+
+        @extend_schema(
+            summary=_('Get a list of segment IDs'),
+            description=_('Get a list of segment IDs'),
+            operation_id=_('Get a list of segment IDs'),
+            parameters=DocumentSplitPatternAPI.get_parameters(),
+            responses=DocumentSplitPatternAPI.get_response(),
+            tags=[_('Knowledge Base/Documentation')]
+        )
+        def get(self, request: Request, workspace_id: str, knowledge_id: str):
+            return result.success(DocumentSerializers.SplitPattern(
+                data={'knowledge_id': knowledge_id, 'workspace_id': workspace_id}
+            ).list())
+
     class BatchEditHitHandling(APIView):
         authentication_classes = [TokenAuth]