我正在使用pandas.read_sql_query函数读取一些sql文件。一个查询在我挑选出来的某个特定位上抛出错误。
（python位--没有什么特别之处，可用于其他查询）

@contextmanager
def open_db_connection(connection_string):
    pyodbc.pooling = False
    connection = pyodbc.connect(connection_string)

    try:
        yield connection
    except pyodbc.DatabaseError as err:
        error, = err.args
        sys.stderr.write(error.message)
    finally:
        connection.close()

noCount = """ SET NOCOUNT ON; """
with open_db_connection(connection_string) as conn:
    res = pd.read_sql_query(noCount+queryObj, conn)

下面的sql语句抛出了一个错误，我不知道为什么会这样。前面的语句和各种临时表都可以工作，可以用pandas.read_sql_query（）收集，但是在下面的位它就坏了。

IF OBJECT_ID('tempdb..#test1') IS NOT NULL DROP TABLE #test1;
select t.PositionID, b.SecurityID
into #test1
from #tmp as t
inner join placeholder.dbo.items  as b on (b.PositionID = t.PositionID and b.StudyDate = '20191230')
where t.ast = 'eq';

IF OBJECT_ID('tempdb..#test2') IS NOT NULL DROP TABLE #test2;
select t.PositionID,    
       case
       when count(i.beta_index)=0 then 1
       else count(i.beta_index)
       end as noIndex
into #test2 
from #test1 as t 
   left join #beta_index as i on (t.SecurityID = i.isin)
   group by t.PositionID;

select * from #test2

这应该会从test 2返回数据。不过需要注意的是-它在SQL Server Management Studio中执行和运行得非常好。