Authenticating on Web.py - will this code be unsafe for production?

I am making a simple web-app which requires login for the admin page. I came across this incantation on the web.py site (http://webpy.org/cookbook/userauth) :

import hashlib
import web    

def POST(self):
    i = web.input()

    authdb = sqlite3.connect('users.db')
    pwdhash = hashlib.md5(i.pa开发者_运维百科ssword).hexdigest()
    check = authdb.execute('select * from users where username=? and password=?', (i.username, pwdhash))
    if check: 
        session.loggedin = True
        session.username = i.username
        raise web.seeother('/results')   
    else: return render.base("Those login details don't work.")

However the page also gives a somewhat ominous warning: "Do not use this code on real site - this is only for illustration.". I was wondering if there are any major holes in this, I'm somewhat unfamiliar with web-programming so just wanted to make sure that using this code wont unwittingly make the app open to trivial attack vectors?

Many thanks

---

select * from users where username=? and password=?', (i.username, pwdhash)

^ SQL injection, broseph. If someone does 'or 1=1' into the search field, they'll get the first result in users because of the SELECT * from. Often the first entry is the admin credentials.

---

The only glaringly obvious problem I see is that the password is stored with as simple MD5 hash with no salt. From your point of view, this isn't so much of an issue, but from the user's point of view it's a major security flaw since someone with access to the database can fairly easily crack sufficiently bad passwords by just googling their MD5 hashes.

---

The only possible problem I can think of here, could be if it would somehow be possible to exploit MD5 collisions, i.e. that two different strings can generate the same MD5 hash - in that case someone could potentially log in with a password that is not correct, but generates the same MD5 hash.

Changing to a better hashing algorithm such as SHA-1 (or something else available in hashlib) would close this potential security problem.

As far as I know, it would be very difficult to exploit the MD5 collision problem to gain access. Even so, it is broken, and quoting security guru Bruce Schneier from the wikipedia article:

[he] wrote of the attack that "[w]e already knew that MD5 is a broken hash function" and that "no one should be using MD5 anymore."