fix: chat bugs (#3308)

shaohuzhang1 · web-flow · commit 598b72fd122e · 2025-06-19T14:53:24.000+08:00
diff --git a/apps/application/chat_pipeline/step/chat_step/i_chat_step.py b/apps/application/chat_pipeline/step/chat_step/i_chat_step.py
@@ -75,7 +75,7 @@ class InstanceSerializer(serializers.Serializer):
         no_references_setting = NoReferencesSetting(required=True,
                                                     label=_("No reference segment settings"))
 
-        user_id = serializers.UUIDField(required=True, label=_("User ID"))
+        workspace_id = serializers.CharField(required=True, label=_("Workspace ID"))
 
         model_setting = serializers.DictField(required=True, allow_null=True,
                                               label=_("Model settings"))
@@ -102,7 +102,7 @@ def execute(self, message_list: List[BaseMessage],
                 chat_id, problem_text,
                 post_response_handler: PostResponseHandler,
                 model_id: str = None,
-                user_id: str = None,
+                workspace_id: str = None,
                 paragraph_list=None,
                 manage: PipelineManage = None,
                 padding_problem_text: str = None, stream: bool = True, chat_user_id=None, chat_user_type=None,
diff --git a/apps/application/chat_pipeline/step/chat_step/impl/base_chat_step.py b/apps/application/chat_pipeline/step/chat_step/impl/base_chat_step.py
@@ -26,7 +26,7 @@
 from application.chat_pipeline.step.chat_step.i_chat_step import IChatStep, PostResponseHandler
 from application.flow.tools import Reasoning
 from application.models import ApplicationChatUserStats, ChatUserType
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 def add_access_num(chat_user_id=None, chat_user_type=None, application_id=None):
@@ -157,7 +157,7 @@ def execute(self, message_list: List[BaseMessage],
                 problem_text,
                 post_response_handler: PostResponseHandler,
                 model_id: str = None,
-                user_id: str = None,
+                workspace_id: str = None,
                 paragraph_list=None,
                 manage: PipelineManage = None,
                 padding_problem_text: str = None,
@@ -167,8 +167,8 @@ def execute(self, message_list: List[BaseMessage],
                 model_params_setting=None,
                 model_setting=None,
                 **kwargs):
-        chat_model = get_model_instance_by_model_user_id(model_id, user_id,
-                                                         **model_params_setting) if model_id is not None else None
+        chat_model = get_model_instance_by_model_workspace_id(model_id, workspace_id,
+                                                              **model_params_setting) if model_id is not None else None
         if stream:
             return self.execute_stream(message_list, chat_id, problem_text, post_response_handler, chat_model,
                                        paragraph_list,
diff --git a/apps/application/chat_pipeline/step/reset_problem_step/i_reset_problem_step.py b/apps/application/chat_pipeline/step/reset_problem_step/i_reset_problem_step.py
@@ -27,7 +27,7 @@ class InstanceSerializer(serializers.Serializer):
                                                     label=_("History Questions"))
         # 大语言模型
         model_id = serializers.UUIDField(required=False, allow_null=True, label=_("Model id"))
-        user_id = serializers.UUIDField(required=True, label=_("User ID"))
+        workspace_id = serializers.CharField(required=True, label=_("User ID"))
         problem_optimization_prompt = serializers.CharField(required=False, max_length=102400,
                                                             label=_("Question completion prompt"))
 
@@ -50,6 +50,6 @@ def _run(self, manage: PipelineManage):
     @abstractmethod
     def execute(self, problem_text: str, history_chat_record: List[ChatRecord] = None, model_id: str = None,
                 problem_optimization_prompt=None,
-                user_id=None,
+                workspace_id=None,
                 **kwargs):
         pass
diff --git a/apps/application/chat_pipeline/step/reset_problem_step/impl/base_reset_problem_step.py b/apps/application/chat_pipeline/step/reset_problem_step/impl/base_reset_problem_step.py
@@ -14,7 +14,7 @@
 from application.chat_pipeline.step.reset_problem_step.i_reset_problem_step import IResetProblemStep
 from application.models import ChatRecord
 from common.utils.split_model import flat_map
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 prompt = _(
     "() contains the user's question. Answer the guessed user's question based on the context ({question}) Requirement: Output a complete question and put it in the <data></data> tag")
@@ -23,9 +23,9 @@
 class BaseResetProblemStep(IResetProblemStep):
     def execute(self, problem_text: str, history_chat_record: List[ChatRecord] = None, model_id: str = None,
                 problem_optimization_prompt=None,
-                user_id=None,
+                workspace_id=None,
                 **kwargs) -> str:
-        chat_model = get_model_instance_by_model_user_id(model_id, user_id) if model_id is not None else None
+        chat_model = get_model_instance_by_model_workspace_id(model_id, workspace_id) if model_id is not None else None
         if chat_model is None:
             return problem_text
         start_index = len(history_chat_record) - 3
diff --git a/apps/application/chat_pipeline/step/search_dataset_step/i_search_dataset_step.py b/apps/application/chat_pipeline/step/search_dataset_step/i_search_dataset_step.py
@@ -44,7 +44,7 @@ class InstanceSerializer(serializers.Serializer):
             validators.RegexValidator(regex=re.compile("^embedding|keywords|blend$"),
                                       message=_("The type only supports embedding|keywords|blend"), code=500)
         ], label=_("Retrieval Mode"))
-        user_id = serializers.UUIDField(required=True, label=_("User ID"))
+        workspace_id = serializers.CharField(required=True, label=_("Workspace ID"))
 
     def get_step_serializer(self, manage: PipelineManage) -> Type[InstanceSerializer]:
         return self.InstanceSerializer
@@ -58,19 +58,19 @@ def _run(self, manage: PipelineManage):
     def execute(self, problem_text: str, knowledge_id_list: list[str], exclude_document_id_list: list[str],
                 exclude_paragraph_id_list: list[str], top_n: int, similarity: float, padding_problem_text: str = None,
                 search_mode: str = None,
-                user_id=None,
+                workspace_id=None,
                 **kwargs) -> List[ParagraphPipelineModel]:
         """
         关于 用户和补全问题 说明: 补全问题如果有就使用补全问题去查询 反之就用用户原始问题查询
         :param similarity:                         相关性
         :param top_n:                              查询多少条
         :param problem_text:                       用户问题
-        :param knowledge_id_list:                    需要查询的数据集id列表
+        :param knowledge_id_list:                  需要查询的数据集id列表
         :param exclude_document_id_list:           需要排除的文档id
         :param exclude_paragraph_id_list:          需要排除段落id
         :param padding_problem_text                补全问题
         :param search_mode                         检索模式
-        :param user_id                             用户id
+        :param workspace_id                        工作空间id
         :return: 段落列表
         """
         pass
diff --git a/apps/application/chat_pipeline/step/search_dataset_step/impl/base_search_dataset_step.py b/apps/application/chat_pipeline/step/search_dataset_step/impl/base_search_dataset_step.py
@@ -25,13 +25,13 @@
 from models_provider.tools import get_model
 
 
-def get_model_by_id(_id, user_id):
-    model = QuerySet(Model).filter(id=_id).first()
+def get_model_by_id(_id, workspace_id):
+    model = QuerySet(Model).filter(id=_id, model_type="EMBEDDING").first()
     if model is None:
         raise Exception(_("Model does not exist"))
-    if model.permission_type == 'PRIVATE' and str(model.user_id) != str(user_id):
-        message = lazy_format(_('No permission to use this model {model_name}'), model_name=model.name)
-        raise Exception(message)
+    if model.workspace_id is not None:
+        if model.workspace_id != workspace_id:
+            raise Exception(_("Model does not exist"))
     return model
 
 
@@ -50,13 +50,13 @@ class BaseSearchDatasetStep(ISearchDatasetStep):
     def execute(self, problem_text: str, knowledge_id_list: list[str], exclude_document_id_list: list[str],
                 exclude_paragraph_id_list: list[str], top_n: int, similarity: float, padding_problem_text: str = None,
                 search_mode: str = None,
-                user_id=None,
+                workspace_id=None,
                 **kwargs) -> List[ParagraphPipelineModel]:
         if len(knowledge_id_list) == 0:
             return []
         exec_problem_text = padding_problem_text if padding_problem_text is not None else problem_text
         model_id = get_embedding_id(knowledge_id_list)
-        model = get_model_by_id(model_id, user_id)
+        model = get_model_by_id(model_id, workspace_id)
         self.context['model_name'] = model.name
         embedding_model = ModelManage.get_model(model_id, lambda _id: get_model(model))
         embedding_value = embedding_model.embed_query(exec_problem_text)
diff --git a/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py b/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py
@@ -11,7 +11,6 @@
 import re
 import time
 from functools import reduce
-from types import AsyncGeneratorType
 from typing import List, Dict
 
 from django.db.models import QuerySet
@@ -24,7 +23,7 @@
 from application.flow.step_node.ai_chat_step_node.i_chat_node import IChatNode
 from application.flow.tools import Reasoning
 from models_provider.models import Model
-from models_provider.tools import get_model_credential, get_model_instance_by_model_user_id
+from models_provider.tools import get_model_credential, get_model_instance_by_model_workspace_id
 
 tool_message_template = """
 <details>
@@ -206,8 +205,9 @@ def execute(self, model_id, system, prompt, dialogue_number, history_chat_record
             model_setting = {'reasoning_content_enable': False, 'reasoning_content_end': '</think>',
                              'reasoning_content_start': '<think>'}
         self.context['model_setting'] = model_setting
-        chat_model = get_model_instance_by_model_user_id(model_id, self.flow_params_serializer.data.get('user_id'),
-                                                         **model_params_setting)
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        chat_model = get_model_instance_by_model_workspace_id(model_id, workspace_id,
+                                                              **model_params_setting)
         history_message = self.get_history_message(history_chat_record, dialogue_number, dialogue_type,
                                                    self.runtime_node_id)
         self.context['history_message'] = history_message
diff --git a/apps/application/flow/step_node/image_generate_step_node/impl/base_image_generate_node.py b/apps/application/flow/step_node/image_generate_step_node/impl/base_image_generate_node.py
@@ -9,7 +9,7 @@
 from application.flow.step_node.image_generate_step_node.i_image_generate_node import IImageGenerateNode
 from common.utils.common import bytes_to_uploaded_file
 from oss.serializers.file import FileSerializer
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 class BaseImageGenerateNode(IImageGenerateNode):
@@ -25,8 +25,9 @@ def execute(self, model_id, prompt, negative_prompt, dialogue_number, dialogue_t
                 **kwargs) -> NodeResult:
         print(model_params_setting)
         application = self.workflow_manage.work_flow_post_handler.chat_info.application
-        tti_model = get_model_instance_by_model_user_id(model_id, self.flow_params_serializer.data.get('user_id'),
-                                                        **model_params_setting)
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        tti_model = get_model_instance_by_model_workspace_id(model_id, workspace_id,
+                                                             **model_params_setting)
         history_message = self.get_history_message(history_chat_record, dialogue_number)
         self.context['history_message'] = history_message
         question = self.generate_prompt_question(prompt)
diff --git a/apps/application/flow/step_node/image_understand_step_node/impl/base_image_understand_node.py b/apps/application/flow/step_node/image_understand_step_node/impl/base_image_understand_node.py
@@ -11,7 +11,7 @@
 from application.flow.i_step_node import NodeResult, INode
 from application.flow.step_node.image_understand_step_node.i_image_understand_node import IImageUnderstandNode
 from knowledge.models import File
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 def _write_context(node_variable: Dict, workflow_variable: Dict, node: INode, workflow, answer: str):
@@ -79,9 +79,9 @@ def execute(self, model_id, system, prompt, dialogue_number, dialogue_type, hist
         # 处理不正确的参数
         if image is None or not isinstance(image, list):
             image = []
-        print(model_params_setting)
-        image_model = get_model_instance_by_model_user_id(model_id, self.flow_params_serializer.data.get('user_id'),
-                                                          **model_params_setting)
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        image_model = get_model_instance_by_model_workspace_id(model_id, workspace_id,
+                                                               **model_params_setting)
         # 执行详情中的历史消息不需要图片内容
         history_message = self.get_history_message_for_details(history_chat_record, dialogue_number)
         self.context['history_message'] = history_message
diff --git a/apps/application/flow/step_node/question_node/impl/base_question_node.py b/apps/application/flow/step_node/question_node/impl/base_question_node.py
@@ -18,7 +18,7 @@
 from application.flow.i_step_node import NodeResult, INode
 from application.flow.step_node.question_node.i_question_node import IQuestionNode
 from models_provider.models import Model
-from models_provider.tools import get_model_instance_by_model_user_id, get_model_credential
+from models_provider.tools import get_model_instance_by_model_workspace_id, get_model_credential
 
 
 def _write_context(node_variable: Dict, workflow_variable: Dict, node: INode, workflow, answer: str):
@@ -87,8 +87,9 @@ def execute(self, model_id, system, prompt, dialogue_number, history_chat_record
                 **kwargs) -> NodeResult:
         if model_params_setting is None:
             model_params_setting = get_default_model_params_setting(model_id)
-        chat_model = get_model_instance_by_model_user_id(model_id, self.flow_params_serializer.data.get('user_id'),
-                                                         **model_params_setting)
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        chat_model = get_model_instance_by_model_workspace_id(model_id, workspace_id,
+                                                              **model_params_setting)
         history_message = self.get_history_message(history_chat_record, dialogue_number)
         self.context['history_message'] = history_message
         question = self.generate_prompt_question(prompt)
diff --git a/apps/application/flow/step_node/reranker_node/impl/base_reranker_node.py b/apps/application/flow/step_node/reranker_node/impl/base_reranker_node.py
@@ -12,7 +12,7 @@
 
 from application.flow.i_step_node import NodeResult
 from application.flow.step_node.reranker_node.i_reranker_node import IRerankerNode
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 def merge_reranker_list(reranker_list, result=None):
@@ -78,8 +78,9 @@ def execute(self, question, reranker_setting, reranker_list, reranker_model_id,
         self.context['document_list'] = [{'page_content': document.page_content, 'metadata': document.metadata} for
                                          document in documents]
         self.context['question'] = question
-        reranker_model = get_model_instance_by_model_user_id(reranker_model_id,
-                                                             self.flow_params_serializer.data.get('user_id'),
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        reranker_model = get_model_instance_by_model_workspace_id(reranker_model_id,
+                                                             workspace_id,
                                                              top_n=top_n)
         result = reranker_model.compress_documents(
             documents,
diff --git a/apps/application/flow/step_node/search_dataset_node/impl/base_search_dataset_node.py b/apps/application/flow/step_node/search_dataset_node/impl/base_search_dataset_node.py
@@ -19,7 +19,7 @@
 from common.utils.common import get_file_content
 from knowledge.models import Document, Paragraph, Knowledge, SearchMode
 from maxkb.conf import PROJECT_DIR
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 def get_embedding_id(dataset_id_list):
@@ -67,7 +67,8 @@ def execute(self, dataset_id_list, dataset_setting, question,
         if len(dataset_id_list) == 0:
             return get_none_result(question)
         model_id = get_embedding_id(dataset_id_list)
-        embedding_model = get_model_instance_by_model_user_id(model_id, self.flow_params_serializer.data.get('user_id'))
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        embedding_model = get_model_instance_by_model_workspace_id(model_id, workspace_id)
         embedding_value = embedding_model.embed_query(question)
         vector = VectorStore.get_embedding_vector()
         exclude_document_id_list = [str(document.id) for document in
diff --git a/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py b/apps/application/flow/step_node/speech_to_text_step_node/impl/base_speech_to_text_node.py
@@ -9,7 +9,7 @@
 from application.flow.step_node.speech_to_text_step_node.i_speech_to_text_node import ISpeechToTextNode
 from common.utils.common import split_and_transcribe, any_to_mp3
 from knowledge.models import File
-from models_provider.tools import get_model_instance_by_model_user_id
+from models_provider.tools import get_model_instance_by_model_workspace_id
 
 
 class BaseSpeechToTextNode(ISpeechToTextNode):
@@ -20,7 +20,8 @@ def save_context(self, details, workflow_manage):
             self.answer_text = details.get('answer')
 
     def execute(self, stt_model_id, chat_id, audio, **kwargs) -> NodeResult:
-        stt_model = get_model_instance_by_model_user_id(stt_model_id, self.flow_params_serializer.data.get('user_id'))
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        stt_model = get_model_instance_by_model_workspace_id(stt_model_id, workspace_id)
         audio_list = audio
         self.context['audio_list'] = audio
 
diff --git a/apps/application/flow/step_node/text_to_speech_step_node/impl/base_text_to_speech_node.py b/apps/application/flow/step_node/text_to_speech_step_node/impl/base_text_to_speech_node.py
@@ -6,8 +6,8 @@
 
 from application.flow.i_step_node import NodeResult
 from application.flow.step_node.text_to_speech_step_node.i_text_to_speech_node import ITextToSpeechNode
+from models_provider.tools import get_model_instance_by_model_workspace_id
 from oss.serializers.file import FileSerializer
-from models_provider.tools import get_model_instance_by_model_user_id
 
 
 def bytes_to_uploaded_file(file_bytes, file_name="generated_audio.mp3"):
@@ -42,8 +42,9 @@ def execute(self, tts_model_id, chat_id,
                 content, model_params_setting=None,
                 **kwargs) -> NodeResult:
         self.context['content'] = content
-        model = get_model_instance_by_model_user_id(tts_model_id, self.flow_params_serializer.data.get('user_id'),
-                                                    **model_params_setting)
+        workspace_id = self.workflow_manage.get_body().get('workspace_id')
+        model = get_model_instance_by_model_workspace_id(tts_model_id, workspace_id,
+                                                         **model_params_setting)
         audio_byte = model.text_to_speech(content)
         # 需要把这个音频文件存储到数据库中
         file_name = 'generated_audio.mp3'
diff --git a/apps/application/serializers/application.py b/apps/application/serializers/application.py
diff --git a/apps/application/serializers/common.py b/apps/application/serializers/common.py
diff --git a/apps/application/views/application.py b/apps/application/views/application.py
diff --git a/apps/chat/serializers/chat.py b/apps/chat/serializers/chat.py
diff --git a/apps/models_provider/tools.py b/apps/models_provider/tools.py
