Merge remote-tracking branch 'origin/main'

backend/alembic/versions/016_modify_chat.py

@@ -0,0 +1,71 @@
+"""016_modify_chat
+
+Revision ID: 031148da1d81
+Revises: 02d84523a979
+Create Date: 2025-06-26 17:00:07.054531
+
+"""
+from alembic import op
+import sqlalchemy as sa
+import sqlmodel.sql.sqltypes
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = '031148da1d81'
+down_revision = '02d84523a979'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.alter_column('chat', 'datasource',
+               existing_type=sa.INTEGER(),
+               nullable=True)
+    op.add_column('chat_record', sa.Column('ai_modal_id', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('first_chat', sa.Boolean(), nullable=True))
+    op.add_column('chat_record', sa.Column('recommended_question_answer', sa.Text(), nullable=True))
+    op.add_column('chat_record', sa.Column('recommended_question', sa.Text(), nullable=True))
+    op.add_column('chat_record', sa.Column('datasource_select_answer', sa.Text(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_sql', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_chart', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_analysis', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_predict', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('full_recommended_question_message', sa.Text(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_recommended_question', sa.Integer(), nullable=True))
+    op.add_column('chat_record', sa.Column('full_select_datasource_message', sa.Text(), nullable=True))
+    op.add_column('chat_record', sa.Column('token_select_datasource_question', sa.Integer(), nullable=True))
+    op.alter_column('chat_record', 'chat_id',
+               existing_type=sa.INTEGER(),
+               nullable=False)
+    op.alter_column('chat_record', 'datasource',
+               existing_type=sa.INTEGER(),
+               nullable=True)
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.alter_column('chat_record', 'datasource',
+               existing_type=sa.INTEGER(),
+               nullable=False)
+    op.alter_column('chat_record', 'chat_id',
+               existing_type=sa.INTEGER(),
+               nullable=True)
+    op.drop_column('chat_record', 'token_select_datasource_question')
+    op.drop_column('chat_record', 'full_select_datasource_message')
+    op.drop_column('chat_record', 'token_recommended_question')
+    op.drop_column('chat_record', 'full_recommended_question_message')
+    op.drop_column('chat_record', 'token_predict')
+    op.drop_column('chat_record', 'token_analysis')
+    op.drop_column('chat_record', 'token_chart')
+    op.drop_column('chat_record', 'token_sql')
+    op.drop_column('chat_record', 'datasource_select_answer')
+    op.drop_column('chat_record', 'recommended_question')
+    op.drop_column('chat_record', 'recommended_question_answer')
+    op.drop_column('chat_record', 'first_chat')
+    op.drop_column('chat_record', 'ai_modal_id')
+    op.alter_column('chat', 'datasource',
+               existing_type=sa.INTEGER(),
+               nullable=False)
+    # ### end Alembic commands ###

backend/alembic/versions/017_rsa_ddl.py

@@ -0,0 +1,35 @@
+"""017_rsa_ddl
+
+Revision ID: a0ba8268868d
+Revises: 031148da1d81
+Create Date: 2025-06-27 15:05:38.676825
+
+"""
+from alembic import op
+import sqlalchemy as sa
+import sqlmodel.sql.sqltypes
+
+
+# revision identifiers, used by Alembic.
+revision = 'a0ba8268868d'
+down_revision = '031148da1d81'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    op.create_table(
+        'rsa',
+        sa.Column('id', sa.BigInteger(), primary_key=True, nullable=False),
+        sa.Column('private_key', sa.Text(), default="", nullable=False),
+        sa.Column('public_key', sa.Text(), default="", nullable=False),
+        sa.Column('salt', sa.Text(), default="", nullable=False),
+        sa.Column('create_time', sa.BigInteger(), default=0, nullable=False),
+        sa.Column('update_time', sa.BigInteger(), default=0, nullable=False)
+    )
+    op.create_index(op.f('ix_rsa_id'), 'rsa', ['id'], unique=False)
+
+
+def downgrade():
+    op.drop_table('rsa')
+    op.drop_index(op.f('ix_rsa_id'), table_name='rsa')

backend/apps/chat/api/chat.py

@@ -8,7 +8,7 @@ from sqlmodel import select
 
 from apps.chat.curd.chat import list_chats, get_chat_with_records, create_chat, rename_chat, \
     delete_chat, list_records
-from apps.chat.models.chat_model import CreateChat, ChatRecord, RenameChat, Chat, ChatQuestion
+from apps.chat.models.chat_model import CreateChat, ChatRecord, RenameChat, Chat, ChatQuestion, ChatMcp
 from apps.chat.task.llm import LLMService
 from apps.datasource.crud.datasource import get_table_schema
 from apps.datasource.models.datasource import CoreDatasource
@@ -57,6 +57,18 @@ async def delete(session: SessionDep, chart_id: int):
         )
 
 
+@router.post("/mcp_start", operation_id="mcp_start")
+async def mcp_start(session: SessionDep, chat: ChatMcp):
+    user = await get_current_user(session, chat.token)
+    return create_chat(session, user, CreateChat(), False)
+
+
+@router.post("/mcp_question", operation_id="mcp_question")
+async def mcp_question(session: SessionDep, chat: ChatMcp):
+    user = await get_current_user(session, chat.token)
+    return await stream_sql(session, user, chat)
+
+
 @router.post("/start")
 async def start_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj: CreateChat):
     try:
@@ -68,25 +80,6 @@ async def start_chat(session: SessionDep, current_user: CurrentUser, create_chat
         )
 
 
-@router.post("/mcp_question", operation_id="mcp_question")
-async def mcp_question(session: SessionDep, token: str, request_question: ChatQuestion):
-    user = await get_current_user(session, token)
-    # return await stream_sql(session, user, request_question)
-    return {"content":"""步骤1: 确定需要查询的字段。
-我们需要统计上海的订单总数，因此需要从"城市"字段中筛选出值为"上海"的记录，并使用COUNT函数计算这些记录的数量。
-
-步骤2: 确定筛选条件。
-问题要求统计上海的订单总数，所以我们需要在SQL语句中添加WHERE "城市" = '上海'来筛选出符合条件的记录。
-
-步骤3: 避免关键字冲突。
-因为这个Excel/CSV数据库是 PostgreSQL 类型，所以在schema、表名、字段名和别名外层加双引号。
-
-最终答案:
-```json
-{"success":true,"sql":"SELECT COUNT(*) AS \"TotalOrders\" FROM \"public\".\"Sheet1_c27345b66e\" WHERE \"城市\" = '上海';"}
-```"""}
-
-
 @router.post("/question", operation_id="question")
 async def stream_sql(session: SessionDep, current_user: CurrentUser, request_question: ChatQuestion):
     """Stream SQL analysis results
@@ -106,16 +99,17 @@ async def stream_sql(session: SessionDep, current_user: CurrentUser, request_que
             status_code=400,
             detail=f"Chat with id {request_question.chat_id} not found"
         )
+    ds: CoreDatasource | None = None
+    if chat.datasource:
+        # Get available datasource
+        ds = session.query(CoreDatasource).filter(CoreDatasource.id == chat.datasource).first()
+        if not ds:
+            raise HTTPException(
+                status_code=500,
+                detail="No available datasource configuration found"
+            )
 
-    # Get available datasource
-    ds = session.query(CoreDatasource).filter(CoreDatasource.id == chat.datasource).first()
-    if not ds:
-        raise HTTPException(
-            status_code=500,
-            detail="No available datasource configuration found"
-        )
-
-    request_question.engine = ds.type_name if ds.type != 'excel' else 'PostgreSQL'
+        request_question.engine = ds.type_name if ds.type != 'excel' else 'PostgreSQL'
 
     # Get available AI model
     aimodel = session.exec(select(AiModelDetail).where(
@@ -128,14 +122,18 @@ async def stream_sql(session: SessionDep, current_user: CurrentUser, request_que
             detail="No available AI model configuration found"
         )
 
-    history_records: List[ChatRecord] = list_records(session=session, current_user=current_user,
-                                                     chart_id=request_question.chat_id)
+    history_records: List[ChatRecord] = list(filter(lambda r: True if r.first_chat != True else False,
+                                                    list_records(session=session, current_user=current_user,
+                                                                 chart_id=request_question.chat_id)))
     # get schema
-    request_question.db_schema = get_table_schema(session=session, ds=ds)
+    if ds:
+        request_question.db_schema = get_table_schema(session=session, ds=ds)
+
     db_user = get_user_info(session=session, user_id=current_user.id)
     request_question.lang = db_user.language
 
-    llm_service = LLMService(request_question, aimodel, history_records, CoreDatasource(**ds.model_dump()))
+    llm_service = LLMService(request_question, aimodel, history_records,
+                             CoreDatasource(**ds.model_dump()) if ds else None)
 
     llm_service.init_record(session=session, current_user=current_user)
 
@@ -144,6 +142,16 @@ async def stream_sql(session: SessionDep, current_user: CurrentUser, request_que
             # return id
             yield orjson.dumps({'type': 'id', 'id': llm_service.get_record().id}).decode() + '\n\n'
 
+            # select datasource if datasource is none
+            if not ds:
+                ds_res = llm_service.select_datasource(session=session)
+                for chunk in ds_res:
+                    yield orjson.dumps({'content': chunk, 'type': 'datasource-result'}).decode() + '\n\n'
+                yield orjson.dumps({'id': llm_service.ds.id, 'datasource_name': llm_service.ds.name,
+                                    'engine_type': llm_service.ds.type_name, 'type': 'datasource'}).decode() + '\n\n'
+
+                llm_service.chat_question.db_schema = get_table_schema(session=session, ds=llm_service.ds)
+
             # generate sql
             sql_res = llm_service.generate_sql(session=session)
             full_sql_text = ''

backend/apps/chat/curd/chat.py

@@ -60,6 +60,8 @@ def get_chat_with_records(session: SessionDep, chart_id: int, current_user: Curr
         load_only(ChatRecord.id, ChatRecord.chat_id, ChatRecord.create_time, ChatRecord.finish_time,
                   ChatRecord.question, ChatRecord.sql_answer, ChatRecord.sql, ChatRecord.data,
                   ChatRecord.chart_answer, ChatRecord.chart, ChatRecord.analysis, ChatRecord.predict,
+                  ChatRecord.datasource_select_answer, ChatRecord.recommended_question_answer,
+                  ChatRecord.recommended_question,
                   ChatRecord.predict_data, ChatRecord.finish, ChatRecord.error, ChatRecord.run_time)).filter(
         and_(Chat.create_by == current_user.id, ChatRecord.chat_id == chart_id)).order_by(ChatRecord.create_time).all()
 
@@ -74,7 +76,8 @@ def list_records(session: SessionDep, chart_id: int, current_user: CurrentUser)
     return record_list
 
 
-def create_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj: CreateChat, require_datasource: bool = True) -> ChatInfo:
+def create_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj: CreateChat,
+                require_datasource: bool = True) -> ChatInfo:
     if not create_chat_obj.datasource and require_datasource:
         raise Exception("Datasource cannot be None")
 
@@ -84,7 +87,7 @@ def create_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj:
     chat = Chat(create_time=datetime.datetime.now(),
                 create_by=current_user.id,
                 brief=create_chat_obj.question.strip()[:20])
-    ds: CoreDatasource = None
+    ds: CoreDatasource | None = None
     if create_chat_obj.datasource:
         chat.datasource = create_chat_obj.datasource
         ds = session.query(CoreDatasource).filter(CoreDatasource.id == create_chat_obj.datasource).first()
@@ -93,6 +96,8 @@ def create_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj:
             raise Exception(f"Datasource with id {create_chat_obj.datasource} not found")
 
         chat.engine_type = ds.type_name
+    else:
+        chat.engine_type = ''
 
     chat_info = ChatInfo(**chat.model_dump())
 
@@ -102,15 +107,9 @@ def create_chat(session: SessionDep, current_user: CurrentUser, create_chat_obj:
     chat_info.id = chat.id
     session.commit()
 
-    if not create_chat_obj.datasource:
-        # use AI to get ds
-
-        if not ds:
-            raise Exception(f"Datasource with id {create_chat_obj.datasource} not found")
-
-
-    chat_info.datasource_exists = True
-    chat_info.datasource_name = ds.name
+    if ds:
+        chat_info.datasource_exists = True
+        chat_info.datasource_name = ds.name
 
     return chat_info
 
@@ -205,6 +204,30 @@ def save_full_predict_message_and_answer(session: SessionDep, record_id: int, an
     return result
 
 
+def save_full_select_datasource_message_and_answer(session: SessionDep, record_id: int, answer: str,
+                                                   full_message: str, datasource: int = None,
+                                                   engine_type: str = None) -> ChatRecord:
+    if not record_id:
+        raise Exception("Record id cannot be None")
+    record = session.query(ChatRecord).filter(ChatRecord.id == record_id).first()
+    record.full_select_datasource_message = full_message
+    record.datasource_select_answer = answer
+
+    if datasource:
+        record.datasource = datasource
+        record.engine_type = engine_type
+
+    result = ChatRecord(**record.model_dump())
+
+    session.add(record)
+    session.flush()
+    session.refresh(record)
+
+    session.commit()
+
+    return result
+
+
 def save_sql(session: SessionDep, record_id: int, sql: str) -> ChatRecord:
     if not record_id:
         raise Exception("Record id cannot be None")

backend/apps/chat/models/chat_model.py

@@ -7,8 +7,10 @@ from sqlmodel import SQLModel, Field
 
 from apps.template.generate_analysis.generator import get_analysis_template
 from apps.template.generate_chart.generator import get_chart_template
+from apps.template.generate_guess_question.generator import get_guess_question_template
 from apps.template.generate_predict.generator import get_predict_template
 from apps.template.generate_sql.generator import get_sql_template
+from apps.template.select_datasource.generator import get_datasource_template
 
 
 class Chat(SQLModel, table=True):
@@ -25,11 +27,13 @@ class Chat(SQLModel, table=True):
 class ChatRecord(SQLModel, table=True):
     __tablename__ = "chat_record"
     id: Optional[int] = Field(sa_column=Column(Integer, Identity(always=True), primary_key=True))
-    chat_id: int = Field(sa_column=Column(Integer))
+    chat_id: int = Field(sa_column=Column(Integer, nullable=False))
+    ai_modal_id: Optional[int] = Field(sa_column=Column(Integer))
+    first_chat: bool = Field(sa_column=Column(Boolean, nullable=True, default=False))
     create_time: datetime = Field(sa_column=Column(DateTime(timezone=True), nullable=True))
     finish_time: datetime = Field(sa_column=Column(DateTime(timezone=True), nullable=True))
     create_by: int = Field(sa_column=Column(BigInteger, nullable=True))
-    datasource: int = Field(sa_column=Column(Integer, nullable=False))
+    datasource: int = Field(sa_column=Column(Integer, nullable=True))
     engine_type: str = Field(max_length=64)
     question: str = Field(sa_column=Column(Text, nullable=True))
     sql_answer: str = Field(sa_column=Column(Text, nullable=True))
@@ -41,10 +45,21 @@ class ChatRecord(SQLModel, table=True):
     analysis: str = Field(sa_column=Column(Text, nullable=True))
     predict: str = Field(sa_column=Column(Text, nullable=True))
     predict_data: str = Field(sa_column=Column(Text, nullable=True))
+    recommended_question_answer: str = Field(sa_column=Column(Text, nullable=True))
+    recommended_question: str = Field(sa_column=Column(Text, nullable=True))
+    datasource_select_answer: str = Field(sa_column=Column(Text, nullable=True))
     full_sql_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_sql: int = Field(default=0, nullable=True)
     full_chart_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_chart: int = Field(default=0, nullable=True)
     full_analysis_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_analysis: int = Field(default=0, nullable=True)
     full_predict_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_predict: int = Field(default=0, nullable=True)
+    full_recommended_question_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_recommended_question: int = Field(default=0, nullable=True)
+    full_select_datasource_message: str = Field(sa_column=Column(Text, nullable=True))
+    token_select_datasource_question: int = Field(default=0, nullable=True)
     finish: bool = Field(sa_column=Column(Boolean, nullable=True, default=False))
     error: str = Field(sa_column=Column(Text, nullable=True))
     run_time: float = Field(default=0)
@@ -67,7 +82,7 @@ class ChatInfo(BaseModel):
     create_by: int = None
     brief: str = ''
     chat_type: str = "chat"
-    datasource: int = None
+    datasource: Optional[int] = None
     engine_type: str = ''
     datasource_name: str = ''
     datasource_exists: bool = True
@@ -108,7 +123,24 @@ class AiModelQuestion(BaseModel):
     def predict_user_question(self):
         return get_predict_template()['user'].format(fields=self.fields, data=self.data, lang=self.lang)
 
+    def datasource_sys_question(self):
+        return get_datasource_template()['system']
+
+    def datasource_user_question(self, datasource_list: str = "[]"):
+        return get_datasource_template()['user'].format(question=self.question, data=datasource_list, lang=self.lang)
+
+    def datasource_guess_sys_question(self):
+        return get_guess_question_template()['system']
+
+    def datasource_guess_user_question(self, old_questions: str = "[]"):
+        return get_guess_question_template()['user'].format(question=self.question, schema=self.db_schema,
+                                                            old_questions=old_questions, lang=self.lang)
+
 
 class ChatQuestion(AiModelQuestion):
-    question: str
-    chat_id: int
+    question: str = ''
+    chat_id: int = 0
+
+
+class ChatMcp(ChatQuestion):
+    token:str = ''

backend/apps/chat/task/llm.py

@@ -6,12 +6,15 @@ import orjson
 from langchain_community.utilities import SQLDatabase
 from langchain_core.language_models import BaseLLM
 from langchain_core.messages import BaseMessage, SystemMessage, HumanMessage, AIMessage, AIMessageChunk
+from sqlalchemy import select
+from sqlalchemy.orm import load_only
 
 from apps.ai_model.model_factory import LLMConfig, LLMFactory, get_llm_config
 from apps.chat.curd.chat import save_question, save_full_sql_message, save_full_sql_message_and_answer, save_sql, \
     save_error_message, save_sql_exec_data, save_full_chart_message, save_full_chart_message_and_answer, save_chart, \
-    finish_record, save_full_analysis_message_and_answer, save_full_predict_message_and_answer, save_predict_data
-from apps.chat.models.chat_model import ChatQuestion, ChatRecord
+    finish_record, save_full_analysis_message_and_answer, save_full_predict_message_and_answer, save_predict_data, \
+    save_full_select_datasource_message_and_answer
+from apps.chat.models.chat_model import ChatQuestion, ChatRecord, Chat
 from apps.datasource.models.datasource import CoreDatasource
 from apps.db.db import exec_sql
 from apps.system.models.system_model import AiModelDetail
@@ -30,6 +33,7 @@ class LLMService:
     llm: BaseLLM
     sql_message: List[Union[BaseMessage, dict[str, Any]]] = []
     chart_message: List[Union[BaseMessage, dict[str, Any]]] = []
+    history_records: List[ChatRecord] = []
 
     def __init__(self, chat_question: ChatQuestion, aimodel: AiModelDetail, history_records: List[ChatRecord] = [],
                  ds: CoreDatasource = None):
@@ -41,17 +45,22 @@ class LLMService:
         llm_instance = LLMFactory.create_llm(self.config)
         self.llm = llm_instance.llm
 
+        self.history_records = history_records
+
+        self.init_messages()
+
+    def init_messages(self):
         # self.agent_executor = create_react_agent(self.llm)
         last_sql_messages = list(
             filter(lambda r: True if r.full_sql_message is not None and r.full_sql_message.strip() != '' else False,
-                   history_records))
+                   self.history_records))
         last_sql_message_str = "[]" if last_sql_messages is None or len(last_sql_messages) == 0 else last_sql_messages[
             -1].full_sql_message
 
         last_chart_messages = list(
             filter(
                 lambda r: True if r.full_chart_message is not None and r.full_chart_message.strip() != '' else False,
-                history_records))
+                self.history_records))
         last_chart_message_str = "[]" if last_chart_messages is None or len(last_chart_messages) == 0 else \
             last_chart_messages[-1].full_chart_message
 
@@ -216,6 +225,92 @@ class LLMService:
                                                                                       in
                                                                                       predict_msg]).decode())
 
+    def select_datasource(self, session: SessionDep):
+        datasource_msg: List[Union[BaseMessage, dict[str, Any]]] = []
+        datasource_msg.append(SystemMessage(self.chat_question.datasource_sys_question()))
+        _ds_list = session.exec(select(CoreDatasource).options(
+            load_only(CoreDatasource.id, CoreDatasource.name, CoreDatasource.description))).all()
+        _ds_list_dict = []
+        for _ds in _ds_list:
+            _ds_list_dict.append({'id': _ds[0].id, 'name': _ds[0].name, 'description': _ds[0].description})
+        datasource_msg.append(
+            HumanMessage(self.chat_question.datasource_user_question(orjson.dumps(_ds_list_dict).decode())))
+
+        history_msg = []
+        if self.record.full_select_datasource_message and self.record.full_select_datasource_message.strip() != '':
+            history_msg = orjson.loads(self.record.full_select_datasource_message)
+
+        self.record = save_full_select_datasource_message_and_answer(session=session, record_id=self.record.id,
+                                                                     answer='',
+                                                                     full_message=orjson.dumps(history_msg +
+                                                                                               [{'type': msg.type,
+                                                                                                 'content': msg.content}
+                                                                                                for msg
+                                                                                                in
+                                                                                                datasource_msg]).decode())
+        full_text = ''
+        res = self.llm.stream(datasource_msg)
+        for chunk in res:
+            print(chunk)
+            if isinstance(chunk, dict):
+                full_text += chunk['content']
+                yield chunk['content']
+                continue
+            if isinstance(chunk, AIMessageChunk):
+                full_text += chunk.content
+                yield chunk.content
+                continue
+        datasource_msg.append(AIMessage(full_text))
+
+        json_str = extract_nested_json(full_text)
+
+        _error: Exception | None = None
+        _datasource: int | None = None
+        _engine_type: str | None = None
+        try:
+            data = orjson.loads(json_str)
+
+            if data['id'] and data['id'] != 0:
+                _datasource = data['id']
+                _ds = session.query(CoreDatasource).filter(CoreDatasource.id == _datasource).first()
+                if not _ds:
+                    _datasource = None
+                    raise Exception(f"Datasource configuration with id {_datasource} not found")
+                self.ds = CoreDatasource(**_ds.model_dump())
+                self.chat_question.engine = _ds.type_name if _ds.type != 'excel' else 'PostgreSQL'
+                _engine_type = self.chat_question.engine
+                # save chat
+                _chat = session.query(Chat).filter(Chat.id == self.record.chat_id).first()
+                _chat.datasource = _datasource
+                _chat.engine_type = _ds.type_name
+
+                session.add(_chat)
+                session.flush()
+                session.refresh(_chat)
+                session.commit()
+
+            elif data['fail']:
+                raise Exception(data['fail'])
+            else:
+                raise Exception('No available datasource configuration found')
+
+        except Exception as e:
+            _error = e
+
+        self.record = save_full_select_datasource_message_and_answer(session=session, record_id=self.record.id,
+                                                                     answer=full_text, datasource=_datasource,
+                                                                     engine_type=_engine_type,
+                                                                     full_message=orjson.dumps(history_msg +
+                                                                                               [{'type': msg.type,
+                                                                                                 'content': msg.content}
+                                                                                                for msg
+                                                                                                in
+                                                                                                datasource_msg]).decode())
+        self.init_messages()
+
+        if _error:
+            raise _error
+
     def generate_sql(self, session: SessionDep):
         # append current question
         self.sql_message.append(HumanMessage(self.chat_question.sql_user_question()))

backend/apps/template/generate_guess_question/generator.py

@@ -0,0 +1,6 @@
+from apps.template.template import get_base_template
+
+
+def get_guess_question_template():
+    template = get_base_template()
+    return template['template']['guess']

backend/apps/template/select_datasource/generator.py

@@ -0,0 +1,6 @@
+from apps.template.template import get_base_template
+
+
+def get_datasource_template():
+    template = get_base_template()
+    return template['template']['datasource']

backend/common/utils/whitelist.py

@@ -25,7 +25,10 @@ wlist = [
     "/mcp/message",
     "/datasource/list",
     "/system/aimodel/list",
-    "/chat/mcp_question"
+    "/chat/mcp_question",
+    "/chat/mcp_start",
+    "/system/license",
+    "/system/config/key"
 ]
 
 class WhitelistChecker:

backend/main.py

@@ -1,10 +1,7 @@
 import logging
 from fastapi.concurrency import asynccontextmanager
-from fastapi.responses import FileResponse
-from fastapi.staticfiles import StaticFiles
-import os
 import sentry_sdk
-from fastapi import FastAPI, Path, HTTPException
+from fastapi import FastAPI
 from fastapi.routing import APIRoute
 from starlette.middleware.cors import CORSMiddleware
 from starlette.exceptions import HTTPException as StarletteHTTPException
@@ -15,9 +12,7 @@ from common.core.response_middleware import ResponseMiddleware, exception_handle
 from alembic.config import Config
 from alembic import command
 from fastapi_mcp import FastApiMCP
-
-
-# import sqlbot_xpack
+import sqlbot_xpack
 
 def run_migrations():
     alembic_cfg = Config("alembic.ini")
@@ -53,7 +48,7 @@ mcp = FastApiMCP(
     description="SQLBot MCP Server",
     describe_all_responses=True,
     describe_full_response_schema=True,
-    include_operations=["get_datasource_list", "get_model_list", "mcp_question"]
+    include_operations=["get_datasource_list", "get_model_list", "mcp_question", "mcp_start"]
 )
 
 mcp.mount(mcp_app)
@@ -76,23 +71,11 @@ app.include_router(api_router, prefix=settings.API_V1_STR)
 app.add_exception_handler(StarletteHTTPException, exception_handler.http_exception_handler)
 app.add_exception_handler(Exception, exception_handler.global_exception_handler)
 
-frontend_dist = os.path.abspath("../frontend/dist")
-if not os.path.exists(frontend_dist):
-    logging.warning(f"The front-end build directory does not exist: {frontend_dist}")
-    logging.warning("Please make sure you have built the front-end project")
 
-else:
-
-    @app.get("/", include_in_schema=False)
-    async def read_index():
-        return FileResponse(path=os.path.join(frontend_dist, "index.html"))
-
-
-    app.mount("/", StaticFiles(directory=frontend_dist), name="static")
 
 mcp.setup_server()
 
-# sqlbot_xpack.init_fastapi_app(app)
+sqlbot_xpack.init_fastapi_app(app)
 if __name__ == "__main__":
     import uvicorn
 

backend/pyproject.toml

@@ -35,10 +35,20 @@ dependencies = [
     "oracledb (>=3.1.1,<4.0.0)",
     "pyyaml (>=6.0.2,<7.0.0)",
     "fastapi-mcp (>=0.3.4,<0.4.0)",
+    "sqlbot-xpack==0.0.2",
 ]
 [[tool.uv.index]]
+name = "default"
 url = "https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple"
 default = true
+[[tool.uv.index]]
+
+name = "testpypi"
+url = "https://test.pypi.org/simple"
+explicit = true
+
+[tool.uv.sources]
+sqlbot-xpack = { index = "testpypi" }
 
 [tool.uv]
 dev-dependencies = [

backend/template.yaml

@@ -10,9 +10,9 @@ template:
       - SQL查询的字段名不要自动翻译，别名必须为英文。
       - 计算占比，百分比类型字段，保留两位小数，以%结尾。
       - 生成SQL时，必须避免关键字冲突:
-      - 如{engine}是 PostgreSQL、Oracle，则在schema、表名、字段名、别名外层加双引号；
-      - 如{engine}是 MySQL，则在表名、字段名、别名外层加反引号；
-      - 如{engine}是 Microsoft SQL Server，则在schema、表名、字段名、别名外层加方括号。
+      - 如数据库引擎是 PostgreSQL、Oracle，则在schema、表名、字段名、别名外层加双引号；
+      - 如数据库引擎是 MySQL，则在表名、字段名、别名外层加反引号；
+      - 如数据库引擎是 Microsoft SQL Server，则在schema、表名、字段名、别名外层加方括号。
       - 生成的SQL使用JSON格式返回：
       {{"success":true,"sql":"生成的SQL语句"}}
       - 问题与生成SQL无关时，直接回答：
@@ -94,30 +94,31 @@ template:
   guess:
     system: |
       ### 说明：
-      您的任务是根据给定的表结构和问题，推测用户接下来可能提问的1-4个问题。
+      您的任务是根据给定的表结构，用户问题以及以往用户提问，推测用户接下来可能提问的1-4个问题。
       请遵循以下规则：
-      - 推测的问题不能与用户问题重复。
-      - 推测的问题必须与给出的表结构相关。
+      - 推测的问题不能与当前用户问题重复
+      - 推测的问题必须与给出的表结构相关
+      - 若有以往用户提问列表，则根据以往用户提问列表，推测用户最频繁提问的问题，加入到你生成的推测问题中
       - 生成的推测问题使用JSON格式返回：
-      {["推测问题1", "推测问题2", "推测问题3", "推测问题4"]}
+      ["推测问题1", "推测问题2", "推测问题3", "推测问题4"]
+      - 最多返回4个你推测出的结果
+      - 若无法推测,则返回空数据JSON:
+      []
       
       ### 响应:
-      根据您的指示，这是我生成的与 表结构 和 问题 相关的 JSON：
       ```json
-      
-      ### 请遵守以下输出要求:
-      请逐步推理后给出答案
-      推理过程中不需要输出JSON，仅在最终结果内输出符合要求的JSON
-      步骤1: [思考内容]
-      步骤2: [思考内容]
-      ……
-      最终答案: [结果]
+
     user: |
+      ### 请使用 i18n: {lang} 对应的语言输出你的结果
+      
       ### 表结构:
       {schema}
       
-      ### 问题:
+      ### 当前问题:
       {question}
+      
+      ### 以往提问:
+      {old_questions}
   analysis:
     system: |
       ### 说明：
@@ -148,3 +149,23 @@ template:
       
       ### 数据:
       {data}
+  datasource:
+    system: |
+      ### 说明：
+      你是一个数据分析师，你需要根据用户的提问，以及提供的数据源列表（格式为JSON数组:[{{"id": 数据源ID1,"name":"数据源名称1","description":"数据源描述1"}},{{"id": 数据源ID2,"name":"数据源名称2","description":"数据源描述2"}}]），根据名称和描述找出最符合用户提问的数据源，这个数据源后续将被用来进行数据的分析
+      
+      ### 要求：
+      - 以JSON格式返回你找到的符合提问的数据源ID，格式为：{{"id": 符合要求的数据源ID}}
+      - 如果匹配到多个数据源，则只需要返回其中一个即可
+      - 如果没有符合要求的数据源，则返回：{{"fail":"没有找到匹配的数据源"}}
+      - 不需要思考过程，请直接返回JSON结果
+      
+      ```json
+    user: |
+      ### 请使用 i18n: {lang} 对应的语言输出你的结果
+      
+      ### 数据源列表:
+      {data}
+      
+      ### 问题:
+      {question}

frontend/src/router/watch.ts

@@ -1,12 +1,13 @@
 import { ElMessage } from 'element-plus-secondary'
 import { useCache } from '@/utils/useCache'
 import { useUserStore } from '@/stores/user'
-
+import { request } from '@/utils/request'
 const userStore = useUserStore()
 const { wsCache } = useCache()
 const whiteList = ['/login']
 export const watchRouter = (router: any) => {
   router.beforeEach(async (to: any, from: any, next: any) => {
+    await loadXpackStatic()
     const token = wsCache.get('user.token')
     if (whiteList.includes(to.path)) {
       next()
@@ -32,3 +33,14 @@ export const watchRouter = (router: any) => {
     }
   })
 }
+const loadXpackStatic = () => {
+  if (document.getElementById('sqlbot_xpack_static')) {
+    return Promise.resolve()
+  }
+  const url = '/xpack_static/license-generator.umd.js'
+  return request.loadRemoteScript(url, 'sqlbot_xpack_static', () => {
+    // eslint-disable-next-line @typescript-eslint/ban-ts-comment
+    // @ts-ignore
+    LicenseGenerator?.init(import.meta.env.VITE_API_BASE_URL)
+  })
+}

frontend/src/utils/request.ts

@@ -69,6 +69,15 @@ class HttpService {
         if (token && config.headers) {
           config.headers['X-SQLBOT-TOKEN'] = `Bearer ${token}`
         }
+        if (config.url?.includes('/xpack_static/') && config.baseURL) {
+          config.baseURL = config.baseURL.replace('/api/v1', '')
+          // Skip auth for xpack_static requests
+          return config
+        }
+        // eslint-disable-next-line @typescript-eslint/ban-ts-comment
+        // @ts-ignore
+        const request_key = LicenseGenerator.generate()
+        config.headers['X-SQLBOT-KEY'] = request_key
 
         // Request logging
         // console.log(`[Request] ${config.method?.toUpperCase()} ${config.url}`)
@@ -258,6 +267,38 @@ class HttpService {
       responseType: 'blob',
     })
   }
+
+  public loadRemoteScript(url: string, id?: string, cb?: any): Promise<HTMLElement> {
+    if (!url) {
+      return Promise.reject(new Error('URL is required to load remote script'))
+    }
+    if (id && document.getElementById(id)) {
+      return Promise.resolve(document.getElementById(id) as HTMLElement)
+    }
+    return new Promise<HTMLElement>((resolve, reject) => {
+      this.get(url, {
+        responseType: 'text',
+        headers: {
+          'Content-Type': 'application/javascript',
+        },
+      })
+        .then((response: any) => {
+          const script = document.createElement('script')
+          script.textContent = response
+          script.id = id || `remote-script-${Date.now()}`
+          // Append script to head
+          document.head.appendChild(script)
+          if (cb) {
+            cb()
+          }
+          resolve(script)
+        })
+        .catch((error: any) => {
+          console.error(`Failed to load script from ${url}:`, error)
+          reject(new Error(`Failed to load script from ${url}: ${error.message}`))
+        })
+    })
+  }
 }
 
 // Create singleton instance

frontend/src/views/chat/component/DisplayChartBlock.vue

@@ -63,7 +63,7 @@ defineExpose({
 <template>
   <div v-if="message?.record">
     <div v-if="message.record.chart" class="chart-base-container">
-      <div>
+      <div v-if="data?.length > 0">
         <ChartComponent
           v-if="message.record.id"
           :id="id ?? 'default_chat_id'"
@@ -76,6 +76,7 @@ defineExpose({
           :data="data"
         />
       </div>
+      <el-empty v-else description="No Data" />
     </div>
   </div>
 </template>

frontend/src/views/chat/index.vue

@@ -409,6 +409,8 @@ const sendMessage = async () => {
     currentRecord.error = currentRecord.error + 'Error:' + error
     console.error('Error:', error)
     isTyping.value = false
+  } finally {
+    loading.value = false
   }
 }