ok Merge branch 'main' of github.com:modelscope/ms-agent into release/1.5

Author: wangxingjun778

ms_agent/tools/__init__.py

@@ -1,7 +1,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 from .code import CodeExecutionTool, SandboxManagerFactory
 from .filesystem_tool import FileSystemTool
-from .findata import FinancialDataFetcher
 from .mcp_client import MCPClient
 from .split_task import SplitTask
 from .tool_manager import ToolManager

ms_agent/tools/findata/akshare_source.py

@@ -2,12 +2,12 @@
 import re
 from typing import Any, Dict, List, Optional
 
-import akshare as ak
 import pandas as pd
 from ms_agent.tools.findata.data_source_base import (DataSourceError,
                                                      FinancialDataSource,
                                                      NoDataFoundError)
 from ms_agent.utils import get_logger
+from ms_agent.utils.utils import install_package
 
 logger = get_logger()
 
@@ -25,10 +25,19 @@ class AKShareDataSource(FinancialDataSource):
     """
 
     def __init__(self):
+        logger.info('Installing AKShare package...')
+        try:
+            install_package(package_name='akshare')
+        except Exception as e:
+            raise e
+
+        global akshare
+        import akshare
+
         logger.info('Initializing AKShare data source')
         try:
             # Test AKShare availability
-            ak.tool_trade_date_hist_sina()
+            akshare.tool_trade_date_hist_sina()
             logger.info('AKShare initialized successfully')
         except Exception as e:
             raise DataSourceError(f'Failed to initialize AKShare: {e}')
@@ -141,23 +150,23 @@ def get_historical_k_data(
             if code.startswith('sh.') or code.startswith(
                     'sz.') or code.startswith('bj'):
                 clean_code = self._convert_code(code, market='A')
-                df = ak.stock_zh_a_hist(
+                df = akshare.stock_zh_a_hist(
                     symbol=clean_code,
                     period=period,
                     start_date=st_date,
                     end_date=ed_date,
                     adjust=adjust)
             elif code.startswith('hk'):
                 clean_code = self._convert_code(code, market='HK')
-                df = ak.stock_hk_hist(
+                df = akshare.stock_hk_hist(
                     symbol=clean_code,
                     period=period,
                     start_date=st_date,
                     end_date=ed_date,
                     adjust=adjust)
             else:
                 clean_code = self._convert_code(code, market='US')
-                df = ak.stock_us_hist(
+                df = akshare.stock_us_hist(
                     symbol=clean_code,
                     period=period,
                     start_date=st_date,
@@ -204,7 +213,7 @@ def _get_hk_basic_info(self, code: str) -> pd.DataFrame:
 
         # Try to fetch base info
         try:
-            df_base_info = ak.stock_hk_spot_em()
+            df_base_info = akshare.stock_hk_spot_em()
             stock_info = df_base_info[df_base_info['代码'] == clean_code]
             if not stock_info.empty:
                 df_stock_info = pd.DataFrame({
@@ -220,7 +229,7 @@ def _get_hk_basic_info(self, code: str) -> pd.DataFrame:
 
         # Try to fetch business info
         try:
-            df_business_info = ak.stock_zyjs_ths(symbol=clean_code)
+            df_business_info = akshare.stock_zyjs_ths(symbol=clean_code)
             if not df_business_info.empty:
                 df_business_info = df_business_info.rename(
                     columns={
@@ -256,7 +265,7 @@ def _get_us_basic_info(self, code: str) -> pd.DataFrame:
         symbol = self._convert_code(code, 'US')
 
         try:
-            df = ak.stock_us_spot_em()
+            df = akshare.stock_us_spot_em()
             stock_info = df[df['代码'] == symbol]
 
             if stock_info.empty:
@@ -283,7 +292,7 @@ def _get_a_share_basic_info(self, code: str) -> pd.DataFrame:
         clean_code = self._convert_code(code, 'A')
 
         try:
-            df_base_info = ak.stock_individual_info_em(symbol=clean_code)
+            df_base_info = akshare.stock_individual_info_em(symbol=clean_code)
 
             if df_base_info.empty:
                 raise NoDataFoundError(f'No basic info found for {code}')
@@ -307,7 +316,7 @@ def _get_a_share_basic_info(self, code: str) -> pd.DataFrame:
                 'status': ['1']
             })
 
-            df_business_info = ak.stock_zyjs_ths(symbol=clean_code)
+            df_business_info = akshare.stock_zyjs_ths(symbol=clean_code)
             if df_business_info.empty:
                 raise NoDataFoundError(f'No business info found for {code}')
 
@@ -479,8 +488,9 @@ def _filter_columns(row_df: pd.DataFrame,
         ind_df = pd.DataFrame()
         if code.startswith(('hk.', 'us.')):
             try:
-                ind_df = ak.stock_financial_hk_analysis_indicator_em(symbol=clean_code) if code.startswith('hk.') else \
-                    ak.stock_financial_us_analysis_indicator_em(symbol=clean_code)
+                ind_df = akshare.stock_financial_hk_analysis_indicator_em(
+                    symbol=clean_code) if code.startswith('hk.') else \
+                    akshare.stock_financial_us_analysis_indicator_em(symbol=clean_code)
                 ind_df = _select_row_by_report(ind_df)
             except Exception as e:
                 logger.warning(
@@ -495,7 +505,7 @@ def _filter_columns(row_df: pd.DataFrame,
 
             if needs_indicator:
                 try:
-                    ind_df = ak.stock_financial_analysis_indicator(
+                    ind_df = akshare.stock_financial_analysis_indicator(
                         symbol=clean_code)
                     ind_df = _select_row_by_report(ind_df)
                 except Exception as e:
@@ -517,14 +527,14 @@ def _filter_columns(row_df: pd.DataFrame,
                         continue
 
                     elif data_type == 'balance':
-                        df = ak.stock_balance_sheet_by_report_em(
+                        df = akshare.stock_balance_sheet_by_report_em(
                             symbol=code.replace('.', '').upper())
                         row = _select_row_by_report(df)
                         if not row.empty:
                             result[data_type] = row
 
                     elif data_type == 'cash_flow':
-                        df = ak.stock_cash_flow_sheet_by_report_em(
+                        df = akshare.stock_cash_flow_sheet_by_report_em(
                             symbol=code.replace('.', '').upper())
                         row = _select_row_by_report(df)
                         if not row.empty:
@@ -570,13 +580,13 @@ def get_stock_list(self,
 
         try:
             if data_type == 'sse50':
-                df = ak.index_stock_cons(symbol='000016')
+                df = akshare.index_stock_cons(symbol='000016')
             elif data_type == 'hs300':
-                df = ak.index_stock_cons(symbol='000300')
+                df = akshare.index_stock_cons(symbol='000300')
             elif data_type == 'zz500':
-                df = ak.index_stock_cons(symbol='000905')
+                df = akshare.index_stock_cons(symbol='000905')
             elif data_type == 'all_a_share':
-                df_a_share = ak.stock_zh_a_spot_em()
+                df_a_share = akshare.stock_zh_a_spot_em()
                 df_a_share['market'] = 'A'
                 df = df_a_share[['代码', '名称', 'market']].copy()
                 df = df.rename(columns={'代码': 'code', '名称': 'code_name'})
@@ -593,7 +603,7 @@ def get_trade_dates(self,
         logger.info(f'Fetching trade dates ({start_date} to {end_date})')
 
         try:
-            df = ak.tool_trade_date_hist_sina()
+            df = akshare.tool_trade_date_hist_sina()
 
             # Ensure trade_date is string for comparison
             if 'trade_date' in df.columns:
@@ -633,7 +643,7 @@ def get_macro_data(
         for data_type in data_types:
             try:
                 if data_type in ('deposit_rate', 'loan_rate'):
-                    result[data_type] = ak.rate_interbank()
+                    result[data_type] = akshare.rate_interbank()
                 elif data_type in ('required_reserve_ratio'):
                     raise DataSourceError(
                         'Required reserve ratio is not supported by AKShare')
@@ -661,7 +671,7 @@ def _get_money_supply_data_month(
             start_date: Optional[str] = None,
             end_date: Optional[str] = None) -> pd.DataFrame:
         try:
-            df = ak.macro_china_money_supply()  # from 2008-01 to now
+            df = akshare.macro_china_money_supply()  # from 2008-01 to now
             df['月份'] = pd.to_datetime(df['月份'].str.replace('月份',
                                                            '').str.replace(
                                                                '年', '-'))

ms_agent/tools/findata/baostock_source.py

@@ -1,16 +1,15 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 import threading
-import time
 from contextlib import contextmanager
 from copy import deepcopy
 from typing import Any, Dict, List, Optional
 
-import baostock as bs
 import pandas as pd
 from ms_agent.tools.findata.data_source_base import (DataSourceError,
                                                      FinancialDataSource,
                                                      NoDataFoundError)
 from ms_agent.utils import get_logger
+from ms_agent.utils.utils import install_package
 
 logger = get_logger()
 
@@ -40,7 +39,7 @@ def _cancel_logout(self):
     def _force_logout(self):
         with self._session_lock:
             if self._login_count == 0 and self._is_logged_in:
-                bs.logout()
+                baostock.logout()
                 self._is_logged_in = False
                 logger.debug('BaoStock session closed by idle-timeout')
             self._timer = None
@@ -56,7 +55,7 @@ def ensure_login(self):
         """Ensure BaoStock is logged in (thread-safe)"""
         with self._session_lock:
             if not self._is_logged_in:
-                lg = bs.login()
+                lg = baostock.login()
                 if lg.error_code != '0':
                     raise DataSourceError(
                         f'BaoStock login failed: {lg.error_msg}')
@@ -108,6 +107,15 @@ class BaoStockDataSource(FinancialDataSource):
     ]
 
     def __init__(self):
+        logger.info('Installing BaoStock package...')
+        try:
+            install_package(package_name='baostock')
+        except Exception as e:
+            raise e
+
+        global baostock
+        import baostock
+
         logger.info('Initializing BaoStock data source')
         # Test connection
         with baostock_session():
@@ -146,7 +154,7 @@ def get_historical_k_data(
         logger.info(f'Fetching K-data for {code} ({start_date} to {end_date})')
 
         with baostock_session():
-            rs = bs.query_history_k_data_plus(
+            rs = baostock.query_history_k_data_plus(
                 code=code,
                 fields=fields_str,
                 start_date=start_date,
@@ -160,7 +168,7 @@ def get_stock_basic_info(self, code: str) -> pd.DataFrame:
         logger.info(f'Fetching basic info for {code}')
 
         with baostock_session():
-            rs = bs.query_stock_basic(code=code)
+            rs = baostock.query_stock_basic(code=code)
             return self._query_to_dataframe(rs, f'basic info for {code}')
 
     def get_dividend_data(self,
@@ -171,7 +179,7 @@ def get_dividend_data(self,
         logger.info(f'Fetching dividend data for {code} ({year} {year_type})')
 
         with baostock_session():
-            rs = bs.query_dividend_data(
+            rs = baostock.query_dividend_data(
                 code=code, year=year, yearType=year_type)
             return self._query_to_dataframe(
                 rs, f'dividend data for {code} ({year} {year_type})')
@@ -184,7 +192,7 @@ def get_adjust_factor_data(self, code: str, start_date: str,
         )
 
         with baostock_session():
-            rs = bs.query_adjust_factor(
+            rs = baostock.query_adjust_factor(
                 code=code, start_date=start_date, end_date=end_date)
             return self._query_to_dataframe(
                 rs,
@@ -205,17 +213,17 @@ def get_financial_data(self, code: str, year: str, quarter: int,
         with baostock_session():
             for data_type in data_types:
                 if data_type == 'profit':
-                    query_func = bs.query_profit_data
+                    query_func = baostock.query_profit_data
                 elif data_type == 'operation':
-                    query_func = bs.query_operation_data
+                    query_func = baostock.query_operation_data
                 elif data_type == 'growth':
-                    query_func = bs.query_growth_data
+                    query_func = baostock.query_growth_data
                 elif data_type == 'balance':
-                    query_func = bs.query_balance_data
+                    query_func = baostock.query_balance_data
                 elif data_type == 'cash_flow':
-                    query_func = bs.query_cash_flow_data
+                    query_func = baostock.query_cash_flow_data
                 elif data_type == 'dupont':
-                    query_func = bs.query_dupont_data
+                    query_func = baostock.query_dupont_data
                 else:
                     raise ValueError(f'Invalid data type: {data_type}')
 
@@ -252,10 +260,10 @@ def get_report(self,
 
         with baostock_session():
             if report_type == 'performance_express':
-                rs = bs.query_performance_express_report(
+                rs = baostock.query_performance_express_report(
                     code=code, start_date=start_date, end_date=end_date)
             elif report_type == 'performance_forecast':
-                rs = bs.query_forecast_report(
+                rs = baostock.query_forecast_report(
                     code=code, start_date=start_date, end_date=end_date)
             else:
                 raise ValueError(f'Invalid report type: {report_type}')
@@ -274,7 +282,7 @@ def get_stock_industry(self,
         )
 
         with baostock_session():
-            rs = bs.query_stock_industry(code=code, date=date)
+            rs = baostock.query_stock_industry(code=code, date=date)
             return self._query_to_dataframe(
                 rs, f'stock industry for {code or "all"} ({date or "latest"})')
 
@@ -288,13 +296,13 @@ def get_stock_list(self,
 
         with baostock_session():
             if data_type == 'sse50':
-                rs = bs.query_sz50_stocks(date=date)
+                rs = baostock.query_sz50_stocks(date=date)
             elif data_type == 'hs300':
-                rs = bs.query_hs300_stocks(date=date)
+                rs = baostock.query_hs300_stocks(date=date)
             elif data_type == 'zz500':
-                rs = bs.query_zz500_stocks(date=date)
+                rs = baostock.query_zz500_stocks(date=date)
             elif data_type == 'all_a_share':
-                rs = bs.query_all_stock(day=date)
+                rs = baostock.query_all_stock(day=date)
             else:
                 raise ValueError(f'Invalid data type: {data_type}')
 
@@ -313,7 +321,8 @@ def get_trade_dates(self,
         )
 
         with baostock_session():
-            rs = bs.query_trade_dates(start_date=start_date, end_date=end_date)
+            rs = baostock.query_trade_dates(
+                start_date=start_date, end_date=end_date)
             return self._query_to_dataframe(rs, 'trade dates')
 
     def get_macro_data(
@@ -341,27 +350,27 @@ def get_macro_data(
                     parsed_end_date = end_date
 
                     if data_type == 'deposit_rate':
-                        query_func = bs.query_deposit_rate_data
+                        query_func = baostock.query_deposit_rate_data
 
                     elif data_type == 'loan_rate':
-                        query_func = bs.query_loan_rate_data
+                        query_func = baostock.query_loan_rate_data
 
                     elif data_type == 'required_reserve_ratio':
-                        query_func = bs.query_required_reserve_ratio_data
+                        query_func = baostock.query_required_reserve_ratio_data
                         if extra_kwargs:
                             parsed_extra_kwargs.update(extra_kwargs)
                         if 'yearType' not in parsed_extra_kwargs:
                             parsed_extra_kwargs['yearType'] = '0'
 
                     elif data_type == 'money_supply_month':
-                        query_func = bs.query_money_supply_data_month
+                        query_func = baostock.query_money_supply_data_month
                         parsed_start_date = pd.to_datetime(
                             start_date).strftime('%Y-%m')
                         parsed_end_date = pd.to_datetime(end_date).strftime(
                             '%Y-%m')
 
                     elif data_type == 'money_supply_year':
-                        query_func = bs.query_money_supply_data_year
+                        query_func = baostock.query_money_supply_data_year
                         parsed_start_date = pd.to_datetime(
                             start_date).strftime('%Y')
                         parsed_end_date = pd.to_datetime(end_date).strftime(