An application bases much of its security on knowing who a user of the system is. Authentication is the process of verifying identity, proving that a subject is who it claims to be.
Native vs Web Yosai APIs
In the examples below, you will see use of the Yosai API, such as
You use the
Yosai API to secure non-web applications. When you are working
with web applications, use the
WebYosai API instead.
Powerful Authentication using a Simple API
Most of your interactions with Yosai are based on the currently executing user,
called a Subject. You can easily obtain a handle on your subject instance
within the "Yosai context" of your call stack by invoking the
Yosai.get_current_subject() references a thread-local stack
that is pushed/popped as context is entered and exited:
from yosai import Yosai, UsernamePasswordToken yosai = Yosai(env_var='YOSAI_SETTINGS') with Yosai.context(yosai): subject = Yosai.get_current_subject()
Logging-In and Logging-Out
When you log-in, you are attempting to make a secured system recognize your presence. The system recognizes who you are by verifying your identity. Once your identity is verified, the system considers you present (logged in) by creating a new, verified session for you.
There are two key processes involved with logging-in: - Session Management - Authentication
Logging-in consists of authentication followed by instantiation of a new session, removing a guest session that was created for you earlier.
As you first interact with a system secured with Yosai, Yosai recognizes you as a guest. A guest is granted an anonymous session (a new one is created) during the guest's first interaction with the system. Once authenticated, you are granted elevated access to the system: your status in the system changes from when you were anonymous. The elevation of status involves deleting the anonymous session and creating a new, verified session for you. Once authenticated, you are considered present through the existence of this verified session.
Authentication, with Example
Authentication is the process of verifying identity.
In the following example, we log-in a Subject by performing password-based authentication. This process would raise an AuthenticationException if it were to fail. Yosai features a rich exception hierarchy that offers detailed explanations as to why a login failed. This exception hierarchy helps developers diagnose bugs or customer service issues related to authentication.
When a developer wishes to authenticate a user using a password-based approach,
the first step requires instantiation of an
AuthenticationToken object. The
UsernamePasswordToken is one such kind of token and is used for password-based
UsernamePasswordToken is passed identifying attributes
(username) and credentials (password).
from yosai.core import Yosai, UsernamePasswordToken yosai = Yosai(env_var='YOSAI_SETTINGS') with Yosai.context(yosai): subject = Yosai.get_current_subject() authc_token = UsernamePasswordToken(username='thedude', credentials='letsgobowling') try: subject.login(authc_token) except UnknownAccountException: # insert here except IncorrectCredentialsException: # insert here except LockedAccountException: # insert here except ExcessiveAttemptsException: # insert here except AuthenticationException: # insert here
As you can see, authentication entails a single method call:
subject.login(authc_token). The Subject API requires a single method call to authenticate, regardless of the underlying authentication strategy chosen.
Cryptographic Hashing of Passwords
For password-based authentication, Yosai uses the Passlib library for cryptographic hashing and password verification.
The default hashing scheme chosen for Yosai is bcrypt_sha256. As per Passlib documentation , the bcrypt_sha256 algorithm works as follows:
- First, the password is encoded to UTF-8 if not already encoded.
- Then, the UTF-8 encoded password is run through SHA2-256, generating a 32-byte digest
The 32-byte digest is encoded using base64, resulting in a 44-byte result (including the trailing padding '='): For the example "password", the output from this stage would be: "XohImNooBHFR0OVvjcYpJ3NgPQ1qq73WKhHvch0VQtg=".
Finally, the base64 string is passed on to the underlying bcrypt algorithm as the new password to be hashed.
For more information about Passlib's bcrypt_sha256, you may access its documentation here
When you log-out, the system no longer recognizes your presence nor will it
honor any prior recognition of your identity (you would have to re-authenticate
yourself if you re-engaged the system). When you log-out a user, you are
releasing the identifying state of the user by the application. A Subject is
logged out when the Subject is done interacting with the application by calling:
subject.logout(), relinquishing all identifying information and
invalidating the user's session. If you are logging out in a web app,
the RememberMe cookie will also be deleted.
Automatic Log Out
Automatic log out happens at session expiration. Yosai expires sessions in two ways: idle timeout and absolute time to live timeout. Please consult the Session Management documentation to learn more about timeouts.
Manual Log Out
When you manually log-out, you are explicitly telling the system that your work
is done and you do no wish to continue your current session with the system.
Manual log-out is initiated by a user engaging a log-out operation through a user
interface, such as click a "log-out" or "sign out" button, which would ultimately
logout method in the Subject API:
from yosai.core import UsernamePasswordToken, Yosai with Yosai.context(yosai): subject = Yosai.get_current_subject() subject.logout()
Factors of Authentication
Authentication methodologies involve three factors:
- something the user knows
- something the user has
- something the user is
Authentication methods that depend on more than one factor, known as multi-factor authentication (MFA) methods, are considered stronger fraud deterrents than single-factor methods as they are more difficult to compromise. A bank ATM transaction involves MFA because it requires something the user has -- a bank card -- and it requires something the user knows -- a PIN code.
Yosai features native support for Time-based One Time Passwords (TOTP), the prevailing standard for one-time password authentication.
Authenticating a user configured for TOTP authentication involves two steps:
- Username / Password Authentication
- One-time Password Authentication
This is considered two-factor authentication because a password represents something that a person knows and a one-time password represents something that the user has -- a token generated from a key stored in a secure environment.
Often, service providers offer two-factor authentication like TOTP as an optional feature rather than force it upon every user. Yosai follows this approach. Yosai provides optional 2FA by supporting two methods of authentication:
- Single factor: username/password
- A two-factor authentication mechanism (TOTP)
The One-time Password Authentication Process in Yosai
Authentication always begins with a user first authenticating itself using a username and password combination. As to how Yosai responds to a successful username/password authentication depends on whether the user is configured for two-factor authentication.
Scenario 1: Single-Factor Authentication
If a user is configured for single-factor authentication and username/password is verified, Yosai returns control to the calling application.
Scenario 2: Two-Factor Authentication
If a user is configured for two-factor authentication and username/password is verified, Yosai signals to the calling application to collect 2FA information from its client by raising an AdditionalAuthenticationRequired exception. Furthermore, Yosai will dispatch an MFAChallenger, if one is configured, to send information to the client. One such implementation of an MFAChallenger would be an SMS gateway that messages a client.
Two-Factor Authentication Sequence
A client authenticates itself using a username and password.
Yosai determines whether the user's account is configured for two-factor authentication. If the user successfully authenticates step 1 and is configured for 2FA, Yosai will signal to its caller to collect 2FA information from client by raising an AdditionalAuthenticationRequired exception. Further, if a MFAChallenger is configured, it is dispatched to coordinate OTP token generation for the client.
A client obtains a 6-digit OTP token and sends it to the server during OTP authentication. An OTP token is produced by a hashing algorithm (sha1) that uses as its inputs a secret key that is unique to a user and an incrementing value (a counter).
Yosai obtains from a persisted datastore an encrypted version of the user's secret key and decrypts it.
Yosai generates its own one-time password token with the decrypted key.
Yosai verifies that the server-side generated token matches the token provided by the client.
If the client and server tokens don't match, Yosai raises an IncorrectCredentialsException to signal that authentication has failed due to an incorrect token provided by the user. If tokens match, control is returned to the application calling Yosai.
Yosai allows developers to regulate account authentication for any particular user account by defining a number of maximum allowable authentication attempts. If a developer defines within yosai's authentication settings an account_lock_threshold, account locking is enabled, using account_lock_threshold as the limit.
Assuming account locking is enabled, the moment that the number of failed authentication attempts exceeds the maximum-allowable threshold, Yosai will lock the account, prohibiting subsequent authentication regardless of whether credentials match.
An AccountStoreRealm obtains an account from storage. Prior to authenticating an account, the realm determines whether account regulation is enabled in Yosai settings. If account regulation is enabled, the account's locked attribute is evaluated to determine whether an account is locked (and if so, when).
If the account is locked: A LockedAccountException is raised, including a tuple containing the timestamp of the current authentication attempt and a timestamp of when the account was locked.
If the account is not locked, authentication proceeds.
When authentication fails, Yosai caches when the failed attempt happend. When the total number of failed attempts exceeds the maximum allowable fails, the account is locked in the underlying accountstore of the realm that facilitates locking. Consequently, failed authentication attempts live in cache until the corresponding cache entry expires or is deleted.
Native Support for 'Remember Me' Services
As shown in the example above, Yosai supports "Remember Me" in addition to the normal login process. Yosai makes a very precise distinction between a remembered Subject and an actual authenticated Subject:
A remembered Subject is not anonymous and has a known identity (i.e. subject.identifiers is non-empty). However, this identity is remembered from a previous authentication during a previous session. A subject is considered remembered if subject.is_remembered returns True.
An authenticated Subject is one that has been successfully authenticated (i.e. the login method was called without any exception raised) during the Subject's current session. A subject is considered authenticated if subject.authenticated returns True.
Remembered and authenticated states are mutually exclusive -- a True value for one indicates a False value for the other and vice versa.
Why the Distinction?
The word 'authentication' has a very strong connotation of proof. That is, there is an expected guarantee that the Subject has proven that it is who it claims to be.
When a user is merely remembered by a previous interaction with the application, the state of proof no longer exists. The remembered identity gives the system an idea who that user probably is, yet the system has no guarantees that the remembered Subject represents an expected user. Once the subject is authenticated, the user is no longer considered only remembered because its identity would have been verified during the current session.
So although many parts of the application can still perform user-specific logic based on the remembered identifiers, such as customized views, it should typically never perform highly-sensitive operations until the user has legitimately verified its identity by executing a successful authentication attempt.
For example, a check whether a Subject can access financial information should almost always depend on subject.authenticated rather than subject.is_remembered to guarantee an expected and verified identity.
The following is a fairly common scenario that helps illustrate why the the distinction between remembered and authenticated is important.
Let's say you're using Amazon.com. You've logged-in successfully and have added a few books to your shopping cart. But you have to run off to a meeting, but forget to log out. By the time the meeting is over, it's time to go home and you leave the office.
The next day when you come in to work, you realize you didn't complete your purchase, so you go back to amazon.com. This time, Amazon 'remembers' who you are, greets you by name, and still gives you some personalized book recommendations. To Amazon, subject.is_remembered would return True.
But, what happens if you try to access your account to update your credit card information to make your book purchase? While Amazon 'remembers' you (is_remembered is True), it cannot guarantee that you are in fact you (for example, maybe a co-worker is using your computer).
So before you can perform a sensitive action like updating credit card information, Amazon will force you to login so that they can guarantee your identity. After you login, your identity has been verified and to Amazon, subject.authenticated would now be True.
This scenario happens so frequently for many types of applications, so the functionality is built in to Yosai so that you may leverage it for your own application. Now, whether you use subject.is_remembered or subject.authenticated to customize your views and workflows is up to you, but Yosai will maintain this fundamental state in case you need it.
An Event is emitted to the singleton EventBus when the results of authentication are obtained, indicating whether authentication succeeded or failed (without compromising credentials). If you would like to learn more about Event processing, please refer to the documentation about Event Processing.
Events are communicated using a publish-subscribe paradigm. In the case of
DefaultAuthenticator publishes an event to a channel (an
internal Event Bus). The EventBus relays an event to consumers who have
subscribed to the event's topic. It relays the event by calling the callback
method registered for a consumer, using the event payload as its argument(s).
The following table lists the Authentication-related events and subscriber(s):
- MRA =
- EL =