pymysql 插入numpy数组

#建议尽量使用3个引号内嵌双引号的形式，即""" - “%s” - “”"
使用pymysql插入numpy 数组，创建表时将data字段设为blob,

create table if not exists user(
id int(11) NOT NULL AUTO_INCREMENT,
vector blob not null,
PRIMARY KEY(id),
);
vector = np.array([1,2,3,4,5],dtype=np.float16)

然后将vector插入user表时，如果用如下语句

id = 1
vector = np.asarray(vector,dtype=np.float16).tostring()
statement = """insert into user_feature(id,vector) values("%s","%s")"""%(id,vector)
cursor.execute(statement)
con.commit()

这种情况会出现语法错误1064，需要改成如下：

id = 1
vector = np.asarray(vector,dtype=np.float16).tostring()
statement = """insert into user_feature(id,vector) values("%s","%s")"""
cursor.execute(statement,(id,vector))
con.commit()

如出现问题，可以将statement中values中的值均加引号，且均用%s代替，
如

" **** values('%s','%s','%s') ***** "
""" ***** values('%s','%s','%s') ****** """

等均可.
将数据取出时，可采用：

cursor.execute("select id,vector for user ")
result = cursor.fetchall()
for id,vector in result:
#将vector 从blob转换为numpy float16 
	vector = np.frombuffer(vector,dtype=np.float16)
	#do something