pymssql执行许多插入值非常慢

Question

python-2.7.15、pymssql-2.1.4、SQL_Server-2018、Windows 10 专业版、MS-Office-2016

import time
import csv
import pymssql

db_settings = {
    "host" : "127.0.0.1",
    "port" : "1433",
    "user" : "sa",
    "password" : "********",
    "database" : "testdb",
    "charset" : "utf8"
               }
conn = pymssql.connect(**db_settings)
cursor = conn.cursor()
ff = csv.reader(open('base.csv', 'r'))
sql = """
    BEGIN
        INSERT INTO Base([name], [year], [update], [status], 
    [timeline], [language], [pic]) VALUES (%s, %s, %s, %s, %s, %s, %s)
    END
    """
now=time.strftime("%M:%S")
t = []
for i in ff:
    i = i[1:]
    if "year" in i:
        pass
    else:
        t.append((i[0], i[1], i[3], i[4], i[6], i[5], i[8]))
cursor.executemany(sql, t)
conn.commit()

end=time.strftime("%M:%S")

print(now+","+end)

“base.csv”文件大小为 21.7 MB，30374 行。当我执行上面的代码时，需要 929 秒才能完成。这意味着每秒只有 32.7 行，太慢了。 谁能帮我看看原因？万分感谢。 :-)

Answer 1

我将pymssql中execute_many的时间从30分钟减少到30秒，就像这样。

在 sql 中，您可以一次创建多行插入语句。如下图所示

INSERT (col_name1, col_name2) 
INTO table_name 
VALUES 
(row1_val1, row1_val2), 
(row2_val1, row2_val2) ... 
(row1000_val1, row1000_val2)

我实现了插入函数，该函数获取数据块并修改查询以通过一次执行插入多个值。

def insert(query, data, chunk=999):
    conn = get_connection()
    cursor = conn.cursor()
    query = query.lower()
    insert_q, values_q = query.split('values') # get part with the query and the parameters
    insert_q += 'values' # add values to make sql query correct after split
    for chunk_data in chunks(data, chunk):
        # chunk_data contains list of row parameters
        flat_list = [item for sublist in chunk_data for item in sublist] # we make it flat to use execute later instead execute_many
        chunk_query = insert_q + ','.join([values_q] * len(chunk_data)) # creating the query with multiple values insert
        cursor.execute(chunk_query, tuple(flat_list))
        conn.commit()

chunks

可以这样实现（感谢这个论坛的精彩回复）

def chunks(lst, n):
    for i in range(0, len(lst), n):
        yield lst[i:i + n]

使用示例

insert('INSERT (user_id, name, surname) INTO users VALUES (%s, %s, %s)',
       [(1, 'Jack', 'Kcaj'), (2, 'Andrew', 'Golara')]