2014-04-03

本文通过展示在一个table上进行操作来论述如何使用Python快速安全地操作sqlite3。

userinfo表通过使用下面语句创建：

create table userinfo(name text, email text)

更快地插入数据

在此用time.clock()来计时，看看以下三种方法的速度。

import sqlite3
import time

def create_tables(dbname):  
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.execute('''create table userinfo(name text, email text)''')
    conn.commit()
    cursor.close()
    conn.close()
def drop_tables(dbname):
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.execute('''drop table userinfo''')
    conn.commit()
    cursor.close()
    conn.close()

def insert1():
    users = [('qq','qq@example.com'),
            ('ww','ww@example.com'),
            ('ee','ee@example.com'),
            ('rr','rr@example.com'),
            ('tt','tt@example.com'),
            ('yy','yy@example.com'),
            ('uu','uu@example.com')
            ]
    start = time.clock()
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    for user in users:
        cursor.execute("insert into userinfo(name, email) values(?, ?)", user)
        conn.commit()
    cursor.close()
    conn.close()
    end = time.clock()
    print start, end, end-start

def insert2():
    users = [('qq','qq@example.com'),
            ('ww','ww@example.com'),
            ('ee','ee@example.com'),
            ('rr','rr@example.com'),
            ('tt','tt@example.com'),
            ('yy','yy@example.com'),
            ('uu','uu@example.com')
            ]
    start = time.clock()
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    for user in users:
        cursor.execute("insert into userinfo(name, email) values(?, ?)", user)
    conn.commit()
    cursor.close()
    conn.close()
    end = time.clock()
    print start, end, end-start

def insert3():
    users = [('qq','qq@example.com'),
            ('ww','ww@example.com'),
            ('ee','ee@example.com'),
            ('rr','rr@example.com'),
            ('tt','tt@example.com'),
            ('yy','yy@example.com'),
            ('uu','uu@example.com')
            ]
    start = time.clock()
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.executemany("insert into userinfo(name, email) values(?, ?)", users)
    conn.commit()
    cursor.close()
    conn.close()
    end = time.clock()
    print start, end, end-start

if __name__ == '__main__':
    dbname = 'test.db'
    create_tables(dbname)
    insert1()
    drop_tables(dbname)
    create_tables(dbname)
    insert2()
    drop_tables(dbname)
    create_tables(dbname)
    insert3()
    drop_tables(dbname)

某次运行结果：

4.05223164501e-07 0.531585119557 0.531584714334
0.755963264089 0.867329935942 0.111366671854
1.0324360882 1.12175173111 0.0893156429109

另外一次运行结果：

4.05223164501e-07 0.565988971446 0.565988566223
0.768132520942 0.843723660494 0.0755911395524
1.04367819446 1.13247636739 0.0887981729298

在运行结果中，第三列表示插入数据使用的时间。综合看来，方法insert1()的速度很慢，原因在于每次insert都commit()。

更安全地操作数据库

先上代码：

import sqlite3

def create_tables(dbname):  
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.execute('''create table userinfo(name text, email text)''')
    conn.commit()
    cursor.close()
    conn.close()
    
def drop_tables(dbname):
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.execute('''drop table userinfo''')
    conn.commit()
    cursor.close()
    conn.close()

def insert():
    users = [('qq','qq@example.com'),
            ('ww','ww@example.com'),
            ('ee','ee@example.com'),
            ('rr','rr@example.com'),
            ('tt','tt@example.com'),
            ('yy','yy@example.com'),
            ('uu','uu@example.com')
            ]
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    cursor.executemany("insert into userinfo(name, email) values(?, ?)", users)
    conn.commit()
    cursor.close()
    conn.close()
    
def insecure_select(text):
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    print "select name from userinfo where email='%s'" % text
    for row in cursor.execute("select name from userinfo where email='%s'" % text):
        print row
def secure_select(text):
    conn = sqlite3.connect(dbname)
    cursor = conn.cursor()
    print "select name from userinfo where email='%s'" % text
    for row in cursor.execute("select name from userinfo where email= ? ", (text,)):
        print row
        
if __name__ == '__main__':
    dbname = 'test.db'
    create_tables(dbname)
    insert()
    insecure_select("uu@example.com")
    insecure_select("' or 1=1;--")
    secure_select("uu@example.com")
    secure_select("' or 1=1;--")
    drop_tables(dbname)

运行结果：

select name from userinfo where email='uu@example.com'
(u'uu',)
select name from userinfo where email='' or 1=1;--'
(u'qq',)
(u'ww',)
(u'ee',)
(u'rr',)
(u'tt',)
(u'yy',)
(u'uu',)
select name from userinfo where email='uu@example.com'
(u'uu',)
select name from userinfo where email='' or 1=1;--'

函数insecure_select(text)和secure_select(text)的本意都是根据email获取对应的用户名信息。但是insecure_select(text)的实现容易引起sql注入。

insecure_select("' or 1=1;--")便是一个例子。在insecure_select()中cursor.execute()只有一个参数，即sql语句，这个生成的sql语句如果有问题，还是会照常执行。

secure_select(text)的实现可以防止sql注入，cursor.execute()的第一个参数使用了占位符?表示要被替代的内容，第二个参数指定每个占位符对应的值，在底层实现上，这种方法（至少）转义了特殊字符，可以防止sql注入。

可以在资料[1]中看到一个完整的使用sql注入进行入侵的过程。资料[2]中有python如何操作sqlite的更加详细的内容。

资料

[1] 实况转播SQL注入全过程，让你知道危害有多大

[2] DB-API 2.0 interface for SQLite databases