Merge branch 'v2' into pr@v2@feat_permission_manage

Author: zhanweizhang7

apps/application/serializers/application_chat_record.py

@@ -12,6 +12,8 @@
 
 from django.db import transaction
 from django.db.models import QuerySet
+from django.db.models.aggregates import Max
+from django.utils.translation import gettext_lazy as _, gettext
 from rest_framework import serializers
 from rest_framework.utils.formatting import lazy_format
 
@@ -21,9 +23,6 @@
 from common.exception.app_exception import AppApiException
 from common.utils.common import post
 from knowledge.models import Paragraph, Document, Problem, ProblemParagraphMapping
-
-from django.utils.translation import gettext_lazy as _, gettext
-
 from knowledge.serializers.common import get_embedding_model_id_by_knowledge_id, update_document_char_length
 from knowledge.serializers.paragraph import ParagraphSerializers
 from knowledge.task.embedding import embedding_by_paragraph, embedding_by_paragraph_list
@@ -229,6 +228,11 @@ def post_improve(self, instance: Dict):
             chat_record.improve_paragraph_id_list.append(paragraph.id)
 
         # 批量保存段落和问题映射
+        max_position = Paragraph.objects.filter(document_id=document_id).aggregate(
+            max_position=Max('position')
+        )['max_position']
+        for i, paragraph in enumerate(paragraphs):
+            paragraph.position = max_position + i + 1
         Paragraph.objects.bulk_create(paragraphs)
         ProblemParagraphMapping.objects.bulk_create(problem_paragraph_mappings)
 
@@ -276,11 +280,17 @@ def improve(self, instance: Dict, with_valid=True):
 
         document_id = self.data.get("document_id")
         knowledge_id = self.data.get("knowledge_id")
-        paragraph = Paragraph(id=uuid.uuid1(),
-                              document_id=document_id,
-                              content=instance.get("content"),
-                              knowledge_id=knowledge_id,
-                              title=instance.get("title") if 'title' in instance else '')
+        max_position = Paragraph.objects.filter(document_id=document_id).aggregate(
+            max_position=Max('position')
+        )['max_position'] or 0
+        paragraph = Paragraph(
+            id=uuid.uuid1(),
+            document_id=document_id,
+            content=instance.get("content"),
+            knowledge_id=knowledge_id,
+            title=instance.get("title") if 'title' in instance else '',
+            position=max_position + 1
+        )
         problem_text = instance.get('problem_text') if instance.get(
             'problem_text') is not None else chat_record.problem_text
         problem, _ = QuerySet(Problem).get_or_create(content=problem_text, knowledge_id=knowledge_id)

apps/chat/urls.py

@@ -10,5 +10,6 @@
     path('profile', views.AuthProfile.as_view()),
     path('application/profile', views.ApplicationProfile.as_view()),
     path('chat_message/<str:chat_id>', views.ChatView.as_view()),
-    path('open', views.OpenView.as_view())
+    path('open', views.OpenView.as_view()),
+    path('captcha', views.CaptchaView.as_view(), name='captcha'),
 ]

apps/chat/views/chat.py

@@ -21,6 +21,8 @@
 from common.constants.permission_constants import ChatAuth
 from common.exception.app_exception import AppAuthenticationFailed
 from common.result import result
+from users.api import CaptchaAPI
+from users.serializers.login import CaptchaSerializer
 
 
 class AnonymousAuthentication(APIView):
@@ -122,3 +124,14 @@ def get(self, request: Request):
             data={'application_id': request.auth.application_id,
                   'chat_user_id': request.auth.chat_user_id, 'chat_user_type': request.auth.chat_user_type,
                   'debug': False}).open())
+
+
+class CaptchaView(APIView):
+    @extend_schema(methods=['GET'],
+                   summary=_("Get captcha"),
+                   description=_("Get captcha"),
+                   operation_id=_("Get captcha"),  # type: ignore
+                   tags=[_("User Management")],  # type: ignore
+                   responses=CaptchaAPI.get_response())
+    def get(self, request: Request):
+        return result.success(CaptchaSerializer().generate())

apps/common/auth/handle/impl/user_token.py

@@ -224,7 +224,9 @@ def get_role_list(user,
                        workspace_user_role_mapping in
                        workspace_user_role_mapping_list] + [user.role], version=version)
         else:
-            role_list = [user.role, get_role_permission(RoleConstants.WORKSPACE_MANAGE, 'default')]
+            role_list = [user.role]
+            if user.role == RoleConstants.ADMIN.value.__str__():
+                role_list = [user.role, get_role_permission(RoleConstants.WORKSPACE_MANAGE, 'default')]
             cache.set(key, role_list, version=version)
             return role_list
     return workspace_list

apps/knowledge/migrations/0001_initial.py

@@ -56,7 +56,7 @@ class Migration(migrations.Migration):
                 ('scope', models.CharField(choices=[('SHARED', '共享'), ('WORKSPACE', '工作空间可用')], default='WORKSPACE', max_length=20, verbose_name='可用范围')),
                 ('file_size_limit', models.IntegerField(default=100, verbose_name='文件大小限制')),
                 ('meta', models.JSONField(default=dict, verbose_name='元数据')),
-                ('embedding_model', models.ForeignKey(default=knowledge.models.knowledge.default_model, on_delete=django.db.models.deletion.DO_NOTHING, to='models_provider.model', verbose_name='向量模型')),
+                ('embedding_model', models.ForeignKey(on_delete=django.db.models.deletion.DO_NOTHING, to='models_provider.model', verbose_name='向量模型')),
                 ('user', models.ForeignKey(on_delete=django.db.models.deletion.DO_NOTHING, to='users.user', verbose_name='所属用户')),
             ],
             options={

apps/knowledge/models/knowledge.py

@@ -95,9 +95,6 @@ def default_status_meta():
     return {"state_time": {}}
 
 
-def default_model():
-    # todo : 这里需要从数据库中获取默认的模型
-    return uuid.UUID('42f63a3d-427e-11ef-b3ec-a8a1595801ab')
 
 
 class KnowledgeFolder(MPTTModel, AppModelMixin):
@@ -128,8 +125,7 @@ class Knowledge(AppModelMixin):
     scope = models.CharField(max_length=20, verbose_name='可用范围', choices=KnowledgeScope.choices,
                              default=KnowledgeScope.WORKSPACE)
     folder = models.ForeignKey(KnowledgeFolder, on_delete=models.DO_NOTHING, verbose_name="文件夹id", default='default')
-    embedding_model = models.ForeignKey(Model, on_delete=models.DO_NOTHING, verbose_name="向量模型",
-                                        default=default_model)
+    embedding_model = models.ForeignKey(Model, on_delete=models.DO_NOTHING, verbose_name="向量模型")
     file_size_limit = models.IntegerField(verbose_name="文件大小限制", default=100)
     file_count_limit = models.IntegerField(verbose_name="文件数量限制", default=50)
     meta = models.JSONField(verbose_name="元数据", default=dict)

apps/knowledge/serializers/paragraph.py

@@ -270,7 +270,7 @@ def save(self, instance: Dict, with_valid=True, with_embedding=True):
                 'knowledge_id': knowledge_id,
                 'document_id': document_id,
                 'workspace_id': self.data.get('workspace_id')
-            }).adjust_position(position=instance.get('position', max_position + 1))
+            }).adjust_position(instance.get('position', max_position + 1))
             # 插入問題
             QuerySet(Problem).bulk_create(problem_model_list) if len(problem_model_list) > 0 else None
             # 插入问题关联关系

apps/knowledge/template/csv_template_en.csv

@@ -0,0 +1,5 @@
+Section title (optional), Section content (required，question answer), Question (optional，one per line in the cell)
+MaxKB product introduction,"MaxKB is a knowledge base question-answering system based on the LLM large language model. MaxKB = Max Knowledge Base，aims to become the most powerful brain of the enterprise。Out-of-the-box: supports direct document upload、automatic crawling of online documents、automatic text splitting and vectorization、and good intelligent question-answering interactive experience；Seamless embedding: supports zero-coding and rapid embedding into third-party business systems；Multi-model support: supports docking with mainstream large models，including Ollama local private large models (such as Llama 2、Llama 3、qwen)、Tongyi Qianwen、OpenAI、Azure OpenAI、Kimi、Zhipu AI、iFlytek Spark and Baidu Qianfan large models、etc.","What is MaxKB?
+MaxKB product introduction
+Large language model supported by MaxKB
+MaxKB advantages"

apps/knowledge/template/csv_template_zh.csv

@@ -0,0 +1,8 @@
+分段标题（选填）,分段内容（必填，问题答案））,问题（选填，单元格内一行一个）
+MaxKB产品介绍,"MaxKB 是一款基于 LLM 大语言模型的知识库问答系统。MaxKB = Max Knowledge Base，旨在成为企业的最强大脑。
+开箱即用：支持直接上传文档、自动爬取在线文档，支持文本自动拆分、向量化，智能问答交互体验好；
+无缝嵌入：支持零编码快速嵌入到第三方业务系统；
+多模型支持：支持对接主流的大模型，包括 Ollama 本地私有大模型（如 Llama 2、Llama 3、qwen）、通义千问、OpenAI、Azure OpenAI、Kimi、智谱 AI、讯飞星火和百度千帆大模型等。","MaxKB是什么？
+MaxKB产品介绍
+MaxKB支持的大语言模型
+MaxKB优势"

apps/knowledge/template/csv_template_zh_Hant.csv

@@ -0,0 +1,8 @@
+分段標題（選填）,分段內容（必填，問題答案））,問題（選填，單元格內一行一個）
+MaxKB產品介紹,"MaxKB 是一款基於 LLM 大語言模型的知識庫問答系統。MaxKB = Max Knowledge Base，旨在成為企業的最強大大腦。
+開箱即用：支援直接上傳文檔、自動爬取線上文檔，支援文字自動分割、向量化，智慧問答互動體驗好；
+無縫嵌入：支援零編碼快速嵌入到第三方業務系統；
+多模型支援：支持對接主流的大模型，包括Ollama 本地私有大模型（如Llama 2、Llama 3、qwen）、通義千問、OpenAI、Azure OpenAI、Kimi、智譜AI、訊飛星火和百度千帆大模型等。 ","MaxKB是什麼？
+MaxKB產品介紹
+MaxKB支援的大語言模型
+MaxKB優勢"