Please note: this is not an official Google product.
This is a secure scaffold package designed primarily to work with Google App Engine (although it is not limited to this).
It is built using Python 3 and Flask.
The scaffold provides the following basic security guarantees by default through
a flask app factory found in secure_scaffold/factories.py
. This app will:
- Set assorted security headers (Strict-Transport-Security, X-Frame-Options,
X-XSS-Protection, X-Content-Type-Options, Content-Security-Policy) with
strong default values to help avoid attacks like Cross-Site Scripting (XSS)
and Cross-Site Script Inclusion. See
add_csp_headers
andsettings.CSP_CONFIG
. - Verify XSRF tokens by default on authenticated requests using any verb other
that GET, HEAD, or OPTIONS. See the
secure_scaffold/xsrf.py
for more information.
As this is currently in beta we have a beta pypi package.
This project can be installed via
pip install toaster-secure-scaffold-rc
For your convenience we also have the option to install with various other dependencies for the contrib APIs such as cloud-tasks, Datastore and Firestore.
These can be included with the []
syntax. For example all of them can be installed
via:
pip install toaster-secure-scaffold-rc[datastore,firestore,tasks]
Once installed you can easily begin a new project by running:
secure_scaffold start-project PROJECT_NAME GCLOUD_PROJECT_NAME
More details about this command can be found by running:
secure_scaffold start-project --help
To use the secure scaffold in your app, use our app generator.
from secure_scaffold import factories
app = factories.AppFactory().generate()
This will automatically set all the needed CSP headers.
To enable XSRF protection add the decorator to the endpoints you need it for. This needs to be set after the route decorator e.g.
from secure_scaffold import factories
from secure_scaffold import xsrf
app = factories.AppFactory().generate()
@app.route('/', methods=['GET', 'POST'])
@xsrf.xsrf_protected()
def index():
return 'Hello World!'
We use Flask Sessions for XSRF, for this you should set up a unique Secret Key for your application.
A random one is set in the app factory, but you should overwrite this yourself, see Flask Sessions
Similar to django settings, to enable multiple settings files you need to set an environment variable. Your folder structure should include a settings folder containing your settings files, for example:
my_project/
settings/
__init__.py
base.py
development.py
production.py
You should then set the environment variable (SETTINGS_MODULE) to the settings you require in that environment.
export SETTINGS_MODULE=settings.development
You can then import your settings in your project like this:
from secure_scaffold.config import settings
The Secure Scaffold provides two methods of authentication. One is an in built authentication system relying on Googles OAuth2 system. The alternative is a system relying on IAP
Second generation App Engine systems have had the Users API removed so the majority of the original functionality isn't available. As a result it is now recommended to use separate systems such as Googles OAuth2. The Secure Scaffold provides a wrapper for this based heavily on this guide.
For this to work there is some minimal setup.
The first is to provide an OAuth Client ID. To do this go to the GCloud console and create
an OAuth Client ID here. Ensure you add
the domains you are using to the Authorised JavaScript Origins
and Authorised redirect URIs
.
You can include localhost:5000
in these to enable using this system in development.
Once you have a Client ID
add it to your settings file like so:
AUTH_OAUTH_CLIENT_ID = 'my-client-id'
Once done all that is required is to register the auth blueprint to your project like so:
from secure_scaffold import factories
from secure_scaffold.contrib.users.auth import auth_handler
app = factories.AppFactory().generate()
app.register_blueprint(auth_handler.blueprint)
This creates two enpoints at /auth/login
and /auth/authenticate
. Login provides a frontend
with a Google sign in button. This sends an API request to /auth/authenticate
which validates
the login procedure and returns an endpoint to redirect to. By default this is /
.
If you want to force a user to be logged in to access a URL you can use the provided
requires_login
decorator like so:
from secure_scaffold import factories
from secure_scaffold.contrib.users.auth import auth_handler
app = factories.AppFactory().generate()
app.register_blueprint(auth_handler.blueprint)
@app.route('/')
@auth_handler.requires_login
def private_view():
return "You can't see this unless you are logged in."
If a user is logged in they will be able to see the page. If they are not they will
automatically be redirected to /auth/login
.
The OAuth2 system is built using a class at secure_scaffold.contrib.users.auth.AuthHandler
.
This class is designed to be easy to subclass and override. For instance if you wanted to change the URL which the user is redirected to on logging in you can do it like so:
from secure_scaffold import factories
from secure_scaffold.contrib.users.auth import AuthHandler
app = factories.AppFactory().generate()
class MyAuthHandler(AuthHandler):
redirect_url = '/after-login'
auth_handler = MyAuthHandler()
app.register_blueprint(auth_handler.blueprint)
This is available at secure_scaffold.contrib.appengine.users
. It provides a User
class which has a few useful methods providing the details of the current user.
It also provides requires_auth
and requires_admin
decorators which enforce the need
for authentication and admin rights respectively on the views they are applied to.
These work almost identically to how they do in the first generation App Engine APIs.
To use these you will need to enable IAP on your App Engine instance. This is provides the app with the correct headers for this functionality.
The Secure Scaffold comes with a built in API for both Datastore and Firestore.
This is a partial ORM - it allows you to simply define data models, to create, retrieve, update and delete them (CRUD). However it does not work with relations or nested entities/documents.
To use this API you must create a settings file with a variable called DATABASE_SETTINGS
.
This has to be a dict with two fields, an engine
field and a settings
field.
engine
must reference a module with the appropriate database engine, this allows the code to switch between databases such as firestore and datastore with a single setting change.settings
are the settings that will be passed to the database engine client. As a minimum these should containproject
pointing to the gcloud project.
The settings should something like this:
DATABASE_SETTINGS = {
'engine': 'secure_scaffold.contrib.db.engine.firestore',
'settings': {
'project': 'my-gcloud-project-id'
}
}
The above uses the firestore engine. To use Datastore instead you should replace
secure_scaffold.contrib.db.engine.firestore
with
secure_scaffold.contrib.db.engine.datastore
. There are no other code changes required.
The API can be used like so:
from secure_scaffold.contrib.db import models
class Person(models.Model):
name = models.Field(str, primary=True)
age = models.Field(int)
arms = models.Field(int, default=2)
social_security = models.Field(str, unique=True)
data = models.Field(dict, required=False)
# Create some people.
Person(name='John', age=30, social_security='111-11-1111').save()
Person(name='Jane', age=30, social_security='222-22-2222').save()
janes = Person.get(name='Jane') # Returns a generator which yields all objects with a name of 'Jane'
for jane in janes:
jane.age = 28
jane.save() # Updates Jane to have an age of 28 in the database.
people = Person.get_all() # Gives a generator which will yield all the people
for person in people:
print(person.name) # Prints 'John' and then 'Jane'
print(person.age) # Prints 30 and then 28
print(person.arms) # Prints 2 and then 2
person.delete() # Deletes the entry in the database.
The API operates a basic validation system - you define fields within a model class, each field has a type and some optional args. If the field receives an object of the wrong type it will raise an error.
The Secure Scaffold comes with a system for setting up tasks with Google Cloud Tasks.
This system works by creating a TaskRunner
class instance and registering functions
as Task
objects using a decorator provided by the TaskRunner
instance.
This creates a view in a Flask blueprint
stored in the TaskRunner
instance and adds a delay
method to the registered
function - allowing the function to be run later by the task queue.
For this module to work you must first install google-cloud-tasks
:
pip install google-cloud-tasks
You must also add a CLOUD_TASKS_QUEUE_CONFIG
object to your settings file.
It should look like this:
CLOUD_TASKS_QUEUE_CONFIG = {
'project': 'YOUR GCLOUD PROJECT NAME',
'location': 'YOUR TASKE QUEUE LOCATION',
'queue': 'YOUR TASK QUEUE NAME'
}
A basic example of the task system is as follows:
from flask import request
from secure_scaffold import factories
from secure_scaffold.contrib.cloud_tasks import tasks
app = factories.AppFactory().generate()
task_runner = tasks.TaskRunner('tasks', __name__, url_prefix='/tasks')
app.register_blueprint(task_runner.blueprint)
@task_runner.task(route='/print_task', methods=['POST'])
def print_task():
arg = request.json().get('arg')
print(arg)
return 'OK'
@app.route('/')
def main():
print_task.delay(arg='Hello, World!')
return 'OK'
This sets up a Secure Scaffold app as well as a task runner. It then registers the blueprint of the task runner in the app.
We create a task function, which gets an argument called arg
from
the global request
object and prints it. This is registered as a task
using the task_runner.task
decorator.
Finally we create a flask route called main
which calls the delay
method
of our task.
Behind the scenes what happens is our task_runner
creates a Task
object
containing this function and a delay
method. It then registers the object
as a flask route at /tasks/print_task/
. The Task
objects delay method
makes a request to the cloud tasks API to create a task for this method in
the queue. It takes any arbitrary arguments and keyword arguments and adds
them to the body of this request - making them accessible via the
flask.request
global variable within the task.
secure_scaffold/
- Top level directory
secure_scaffold/contrib
- Contains non-essential but useful libraries.
- Holds several alternatives to App Engine APIs which are no longer available in second generation instances
secure_scaffold/tests
- Tests for the secure scaffold
secure_scaffold/config.py
- Similar to django settings set up
- Looks for the "SETTINGS_MODULE" environment variable to be set
- See Settings Config below on how to use this
secure_scaffold/factories.py
- The main Flask app factory that applies the security defaults
- See App Factory below on how to use this
secure_scaffold/settings.py
- Security settings
- Defines our CSP headers and other specifics
secure_scaffold/xsrf.py
- Defines XSRF decorators to be used with your flask app
- See XSRF below on how to use this
We recommend setting up a virtual env to install dependencies:
virtualenv env --python=python3
source env/bin/activate
pip install -r dev_requirements.txt
There are some extra dependencies required for the development on specific submodules.
These include:
google-cloud-firestore
for development on `secure_scaffold.contrib.db.engine.firestoregoogle-cloud-datastore
for development on `secure_scaffold.contrib.db.engine.datastoregoogle-cloud-tasks
for development on `secure_scaffold.contrib.cloud_tasks.tasks
To run unit tests:
pytest
- Flask