FlaskSimpleAuth 1.8.1

Simple authentication, authorization and parameters for Flask, emphasizing configurability

Flask Simple Auth

Simple authentication, authorization and parameter checks for Flask, controled from Flask configuration and decorators.

Description

Help to manage authentication, authorizations and parameters in a Flask REST application.

Authentication is available through the get_user function. It is performed on demand when the function is called, automatically when checking for permissions in a per-role authorization model, or possibly forced for all/most paths.

The module implements inheriting the web-server authentication, password authentication (HTTP Basic, or HTTP/JSON parameters), simple time-limited authentication tokens, and a fake authentication scheme useful for application testing.

It allows to have a login route to generate authentication tokens. For registration, support functions allow to hash new passwords consistently with password checks.

Authorization can be managed with a simple decorator to declare required permissions on a route (eg a role name), and relies on a supplied function to check whether a user has this role. This approach is enough for basic authorization management, but would be insufficient for realistic applications where users can edit their own data but not those of others.

Parameters expected in the request can be declared, their presence and type checked, and they are added automatically as named parameters to route functions, skipping the burden of checking them in typical REST functions.

Simple Example

The application code below performs authentication, authorization and parameter checks triggered by decorators. There is no clue in the source about what kind of authentication is used, which is the whole point: authentication schemes are managed elsewhere, not explicitely in the application code.

# app is the Flask application…
# user_to_password_fun is a function returning the hashed password for a user.
# user_in_group_fun is a function telling whether a user is in a group.

# initialize module
import FlaskSimpleAuth as fsa
fsa.setConfig(app, user_to_password_fun, user_in_group_fun)

# users belonging to the "patcher" group can patch "whatever/*"
# the function gets 3 arguments: one coming from the path (id)
# and the remaining two coming from request parameters (some, stuff).
# "some" is mandatory, stuff is optional because it has a default.
@app.route("/whatever/<int:id>", methods=["PATCH"])
@fsa.authorize("patcher")
@fsa.parameters()
def patch_whatever(id: int, some: int, stuff: str = "wow"):
    # ok to do it, with parameters id, some & stuff
    return "", 204

Authentication is manage from the application flask configuration with FSA_* (Flask simple authentication) directives:

FSA_TYPE = 'httpd'     # inherit web-serveur authentication
# OR others such as:
FSA_TYPE = 'basic'     # HTTP Basic auth
FSA_TYPE = 'param'     # HTTP parameter auth

Various aspects of the implemented schemes can be configured with other directives, with reasonable defaults provided so that not much is really needed beyond choosing the authentication scheme.

Documentation

Install

Use pip install FlaskSimpleAuth to install the module, or whatever other installation method you prefer.

Features

This simple module allows configurable authentication (FSA_TYPE):

• httpd web-server checked authentication passed in the request.

• basic http basic auth with a function hook for getting the password hash.

• param same with http parameter-provided login/password.

• password tries basic then param.

• token auth uses a signed parameter to authenticate a user in a realm for some limited time. The token can be obtained by actually authenticating with other methods.

• fake parameter-based auth for fast and simple testing the claimed login is coldly trusted…

I have considered Flask HTTPAuth obviously, which provides many options, but I do not want to force their per-route-only model and explicit classes but rather rely on mandatory request hooks and have everything managed from the configuration file to easily switch between schemes, without impact on the application code.

Note that this is intended for a REST API implementation serving a remote application. It does not make much sense to "login" and "logout" to/from a REST API because the point of the API is to serve and collect data to all who deserve it, i.e. are authorized, unlike a web application which is served while the client is on the page which maintains a session and should disappear when disconnected as the web browser page is wiped out. However, there is still a "login" concept which is only dedicated at obtaining an auth token, that the application client needs to update from time to time.

Note that web-oriented flask authentication modules are not really relevant in the REST API context, where the server does not care about presenting login forms for instance.

Initialisation

The module is initialized by calling setConfig with three arguments:

• the Flask application object.
• a function to retrieve the password hash from the user name.
• a function which tells whether a user is in a group or role.

# app is already initialized and configured the Flask application

# return password hash if any, or None
def get_user_password(user):
    return …

# return whether user is in group
def user_in_group(user, group):
    return …

import FlaskSimpleAuth as fsa
fsa.setConfig(app, get_user_password, user_in_group)

Then the module can be used to retrieve the authenticated user with get_user, which raises AuthException on failures. Some path may require to skip authentication, for instance registering a new user.

Three directives impact how and when authentication is performed.

• FSA_TYPE governs the how: httpd, basic, param, password, token… as described below. Default is httpd.

• FSA_ALWAYS tells whether to perform authentication in a before request hook. Default is True. On authentication failures 401 are returned. One in a route function, get_user will always return the authenticated user and cannot fail.

• FSA_SKIP_PATH is a list of regular expression patterns which are matched against the request path for skipping systematic authentication. Default is empty, i.e. authentication is applied for all paths.

• FSA_LAZY tells whether to attempt authentication lazily when checking an authorization through a authorize decorator. Default is True.

Using Authentication, Authorization and Parameter Check

Then all route functions can take advantage of this information to check for authorizations with the authorize decorator, and for parameters with the parameters decorator.

@app.route("/somewhere", methods=["POST"])
@fsa.authorize("posters")
@fsa.parameters()
def post_somewhere(stuff: str, nstuff: int, bstuff: bool = False):
    …

Note that more advanced permissions (eg users can edit themselves) will still require manual permission checks at the beginning of the function.

An opened route for user registration with mandatory parameters could look like that:

# with FSA_SKIP_PATH = (r"/register", …)
@app.route("/register", methods=["POST"])
@fsa.parameters()
def post_register(user: str, password: str):
    if user_already_exists_somewhere(user):
        return f"cannot create {user}", 409
    add_new_user_with_hashed_pass(user, fsa.hash_password(password))
    return "", 201

For token authentication, a token can be created on a path authenticated by one of the other methods. The code for that would be as simple as:

# token creation route
@app.route("/login", methods=["GET"])
def get_login():
    return jsonify(fsa.create_token(get_user())), 200

The client application will return the token as a parameter for authenticating later requests, till it expires.

The main configuration directive is FSA_TYPE which governs authentication methods used by the get_user function, as described in the following sections:

httpd Authentication

Inherit web server supplied authentication through request.remote_user. This is the default.

There are plenty authentication schemes available in a web server such as Apache or Nginx, all of which probably more efficiently implemented than python code, so this should be the preferred option. However, it could require significant configuration effort compared to the application-side approach.

basic Authentication

HTTP Basic password authentication, which rely on the Authorization HTTP header in the request.

See also Password Authentication below for how the password is retrieved and checked.

param Authentication

HTTP parameter or JSON password authentication. User name and password are passed as request parameters.

The following configuration directives are available:

• FSA_PARAM_USER parameter name for the user name. Default is USER.
• FSA_PARAM_PASS parameter name for the password. Default is PASS.

See also Password Authentication below for how the password is retrieved and checked.

password Authentication

Tries basic then param authentication.

token Authentication

Only rely on signed tokens for authentication. A token certifies that a user is authenticated in a realm up to some time limit. The token is authenticated by a signature which is the hash of the payload (realm, user and limit) and a secret hold by the server. The token syntax is: <realm>:<user>:<limit>:<signature>, for instance: kiva:calvin:20210221160258:4ee89cd4cc7afe0a86b26bdce6d11126. The time limit is an easily parsable UTC timestamp YYYYMMDDHHmmSS so that it can be checked easily by the application client.

The following configuration directives are available:

• FSA_TOKEN_REALM realm of token. Default is the simplified lower case application name.
• FKA_TOKEN_NAME name of parameter holding the auth token. Default is auth.
• FSA_TOKEN_SECRET secret string used for signing tokens. Default is a system-generated random string containing 128 bits. This default with only work with itself, as it is not shared across server instances or processes. Set to None to disable tokens.
• FSA_TOKEN_DELAY number of minutes of token validity. Default is 60 minutes.
• FSA_TOKEN_GRACE number of minutes of grace time for token validity. Default is 0 minutes.
• FSA_TOKEN_HASH hash algorithm used to sign the token. Default is blake2s.
• FSA_TOKEN_LENGTH number of hash bytes kept for token signature. Default is 16.

Function create_token(user) creates a token for the user.

Note that token authentication is always attempted unless the secret is empty. Setting FSA_TYPE to token results in only token authentication to be used.

Also note that token authentication is usually much faster than password verification because password checks are designed to be slow so as to hinder password cracking. Another benefit of token is that it avoids sending passwords over and over. The rational option is to use a password scheme to retrieve a token and then to use it till it expires.

fake Authentication

Trust a parameter for authentication claims. Only for local tests, obviously. This is inforced.

The following configuration directive is available:

• FSA_FAKE_LOGIN name of parameter holding the user name. Default is LOGIN.

Password Authentication (param or basic)

For checking passwords the password (salted hash) must be retrieved through get_user_password(user). This function must be provided by the application when the module is initialized.

The following configuration directives are available to configure passlib password checks:

• FSA_PASSWORD_SCHEME password scheme to use for passwords. Default is bcrypt. See passlib documentation for available options.
• FSA_PASSWORD_OPTIONS relevant options (for passlib.CryptContext). Default is {'bcrypt__default_rounds': 4}.

Beware that modern password checking is often pretty expensive in order to thwart password cracking if the hashed passwords are leaked, so that you do not want to have to use that on every request in real life (eg hundreds milliseconds for passlib bcrypt 12 rounds). The above defaults result in manageable password checks of a few milliseconds. Consider enabling tokens to reduce the authentication load on each request.

Function hash_password(pass) computes the password salted digest compatible with the current configuration.

authorize Decorator

The decorator expects a list of identifiers, which are typically names or numbers. When several groups are specified, any will allow the operation to proceed.

# group ids
ADMIN, WRITE, READ = 1, 2, 3

@app.route("/some/place", methods=["POST"])
@fsa.authorize(ADMIN, WRITE)
def post_some_place():
    …

The check will call user_in_group(user, group) function to check whether the authenticated user belongs to any of the authorized groups.

The following configuration directive is available:

• FSA_LAZY allows the authorize decorator to perform the authentication when needed, which mean that the before request hook can be skipped. Default is True.

Note that this simplistic model does is not enough for non-trivial applications, where permissions on objects often depend on the object owner. For those, careful per-operation authorization will still be needed.

parameters Decorator

This decorators translates automatically request parameters (HTTP or JSON) to function parameters, relying on function type annotations to do that.

By default, the decorator guesses whether parameters are mandatory based on provided default values, i.e. they are optional when a default is provided.

The required parameter allows to declare whether all parameters must be set (when True), or whether they are optional (False) in which case None values are passed if no defaults are given.

The allparams parameter makes all request parameters be translated to named function parameters that can be manipulated as such.

@app.route("/thing/<int:tid>", methods=["PATCH"])
@fsa.parameters()
def patch_thing_tid(tid: int, name: str = None, price: float = None):
    if name is not None:
        update_name(tid, name)
    …
    return "", 204

The parameters decorator must be placed after the authorize decorator.

The decorator also accepts positional string arguments. It expects these parameter names and generates a 400 if any is missing from the request, and passes them to function named parameters. The decorator looks for HTTP or JSON parameters.

@app.route("/thing/<int:tid>", methods=["PUT"])
@fsa.parameters("name")
def put_thing_tid(tid, name):
    …

The decorator also accepts named parameters associated to a type. It expects these parameter names and generate a 400 if any is missing from the request, it converts the parameter string value to the expected type, resulting in a 400 again if the type conversion fails, and it passes these to the function as named parameters.

@app.route("/add", methods=["GET"])
@fsa.parameters(a=float, b=float)
def get_add(a, b):
    return str(a + b), 200

Request parameter string values are converted to the target type. For int, base syntax is accepted, i.e. 0x11, 0b10001 and 17 all mean decimal 17. For bool, False is an empty string, 0, False, otherwise the value is True.

A side-effect of the parameters decorator passing of request parameters as named function parameters is that request parameter names must be valid python identifiers, which excludes keywords such as pass, def or for.

Versions

Sources are available on GitHub and packaged on PyPI.

1.8.1

Fix typo in distribution configuration file.

1.8.0

Merge autoparams and parameters decorators into a single parameters decorator. Make it guess optional parameters based on default values. Fix conversion issues with boolean type parameters. Enhance integer type to accept other base syntaxes. Improve documentation to advertise the simple and elegant approach. Implement decorator with functions instead of a class.

1.7.0

Simplify code. Add FSA_ALWAYS configuration directive and move the authentication before request hook logic inside the module. Add FSA_SKIP_PATH to skip authentication for some paths. Update documentation to reflect this simplified model.

1.6.0

Add autoparams decorator with required or optional parameters. Add typed parameters to parameters decorator. Make parameters pass request parameters as named function parameters. Simplify authorize decorator syntax and implementation. Advise authorize then parameters or autoparams decorator order. Improved documentation.

1.5.0

Flask internal tests with a good coverage. Switch to setup.cfg configuration. Add convenient parameters decorator.

1.4.0

Add FSA_LAZY configuration directive. Simplify code. Improve warning on short secrets. Repackage…

1.3.0

Improved documentation. Reduce default token signature length and default token secret. Warn on random or short token secrets.

1.2.0

Add grace time for auth token validity. Some code refactoring.

1.1.0

Add after request module cleanup.

1.0.0

Add authorize decorator. Add password authentication scheme. Improved documentation.

0.9.0

Initial release in beta.

TODO

Features

• better control which schemes are attempted?
• add support for JWT?

Implementation

• should it be an object instead of a flat module?
• expand tests

How not to forget autorizations?

• set a autorization_checked variable to False before the request
• reset it to True when autorization is checked
• check whether it was done and possibly abort after the request