“Invalid parameter type” (numpy.int64) when inserting rows with executemany()

问题

I try to insert bunch of data to database

insert_list = [(1,1,1,1,1,1),(2,2,2,2,2,2),(3,3,3,3,3,3),....] #up to 10000 tuples in this list

conn = pyodbc.connect('DRIVER={FreeTDS};SERVER=xxxxx;DATABASE=xxxx;UID=xx;PWD=xx;TDS_Version=7.0')
cursor = conn.cursor()

sql = "insert into ScanEMAxEMAHistoryDay(SecurityNumber, EMA1, EMA2, CrossType, DayCross, IsLocalMinMax) values (?, ?, ?, ?, ?, ?)"

cursor.executemany(sql, insert_list)

cursor.executemany(sql, insert_list)

pyodbc.ProgrammingError: ('Invalid parameter type. param-index=4 param-type=numpy.int64', 'HY105')

reduce to 100 tuples:

cursor.executemany(sql, insert_list[:100])

cursor.executemany(sql, insert_list[:100])

pyodbc.ProgrammingError: ('Invalid parameter type. param-index=4 param-type=numpy.int64', 'HY105') cursor.executemany(sql, insert_list[:100])

reduce to 5 tuples:

cursor.executemany(sql, insert_list[:5])
conn.commit()

This can insert to database

I have try to:

sql = 'SET GLOBAL max_allowed_packet=50*1024*1024'
cursor.execute(sql)

before excutemany() but it have an error:

pyodbc.ProgrammingError: ('42000', "[42000] [FreeTDS][SQL Server]'GLOBAL' is not a recognized SET option. (195) (SQLExecDirectW)")

How did i solve this.

Thank you.

回答1:

Your problem is not with the volume of data per se, it is that some of your tuples contain numpy.int64 values that cannot be used directly as parameter values for your SQL statement. For example,

a = numpy.array([10, 11, 12], dtype=numpy.int64)
params = (1, 1, a[1], 1, 1, 1)
crsr.execute(sql, params)

will throw

ProgrammingError: ('Invalid parameter type. param-index=2 param-type=numpy.int64', 'HY105')

because the third parameter value is a numpy.int64 element from your numpy array a. Converting that value with int() will avoid the issue:

a = numpy.array([10, 11, 12], dtype=numpy.int64)
params = (1, 1, int(a[1]), 1, 1, 1)
crsr.execute(sql, params)

By the way, the reason that

sql = 'SET GLOBAL max_allowed_packet=50*1024*1024'
cursor.execute(sql)

didn't work is that max_allowed_packet is a MySQL setting that does not have any meaning for Microsoft SQL Server.

来源：https://stackoverflow.com/questions/41973933/invalid-parameter-type-numpy-int64-when-inserting-rows-with-executemany

标签

python

sql-server

numpy

pyodbc

freetds