chalice 0.8.1

Quickstart

In this tutorial, you’ll use the chalice command line utility to create and deploy a basic REST API. First, you’ll need to install chalice. Using a virtualenv is recommended:

$ pip install virtualenv
$ virtualenv ~/.virtualenvs/chalice-demo
$ source ~/.virtualenvs/chalice-demo/bin/activate

Note: make sure you are using python2.7 or python3.6. The chalice CLI as well as the chalice python package will support the versions of python supported by AWS Lambda. Currently, AWS Lambda supports python2.7 and python3.6, so that’s what this project supports. You can ensure you’re creating a virtualenv with python3.6 by running:

# Double check you have python3.6
$ which python3.6
/usr/local/bin/python3.6
$ virtualenv --python $(which python3.6) ~/.virtualenvs/chalice-demo
$ source ~/.virtualenvs/chalice-demo/bin/activate

Next, in your virtualenv, install chalice:

$ pip install chalice

You can verify you have chalice installed by running:

$ chalice --help
Usage: chalice [OPTIONS] COMMAND [ARGS]...
...

$ mkdir ~/.aws
$ cat >> ~/.aws/config
[default]
aws_access_key_id=YOUR_ACCESS_KEY_HERE
aws_secret_access_key=YOUR_SECRET_ACCESS_KEY
region=YOUR_REGION (such as us-west-2, us-west-1, etc)

$ chalice new-project helloworld

$ cd helloworld
$ ls -la
drwxr-xr-x   .chalice
-rw-r--r--   app.py
-rw-r--r--   requirements.txt

from chalice import Chalice

app = Chalice(app_name='helloworld')


@app.route('/')
def index():
    return {'hello': 'world'}

$ chalice deploy
...
Initiating first time deployment...
https://qxea58oupc.execute-api.us-west-2.amazonaws.com/dev/

$ curl https://qxea58oupc.execute-api.us-west-2.amazonaws.com/dev/
{"hello": "world"}

$ http https://qxea58oupc.execute-api.us-west-2.amazonaws.com/dev/
HTTP/1.1 200 OK
Connection: keep-alive
Content-Length: 18
Content-Type: application/json
Date: Mon, 30 May 2016 17:55:50 GMT
X-Cache: Miss from cloudfront

{
    "hello": "world"
}

Tutorial: URL Parameters

Now we’re going to make a few changes to our app.py file that demonstrate additional capabilities provided by the python serverless microframework for AWS.

Our application so far has a single view that allows you to make an HTTP GET request to /. Now let’s suppose we want to capture parts of the URI:

from chalice import Chalice

app = Chalice(app_name='helloworld')

CITIES_TO_STATE = {
    'seattle': 'WA',
    'portland': 'OR',
}


@app.route('/')
def index():
    return {'hello': 'world'}

@app.route('/cities/{city}')
def state_of_city(city):
    return {'state': CITIES_TO_STATE[city]}

In the example above, we’ve now added a state_of_city view that allows a user to specify a city name. The view function takes the city name and returns name of the state the city is in. Notice that the @app.route decorator has a URL pattern of /cities/{city}. This means that the value of {city} is captured and passed to the view function. You can also see that the state_of_city takes a single argument. This argument is the name of the city provided by the user. For example:

GET /cities/seattle   --> state_of_city('seattle')
GET /cities/portland  --> state_of_city('portland')

Now that we’ve updated our app.py file with this new view function, let’s redeploy our application. You can run chalice deploy from the helloworld directory and it will deploy your application:

$ chalice deploy

Let’s try it out. Note the examples below use the http command from the httpie package. You can install this using pip install httpie:

$ http https://endpoint/dev/cities/seattle
HTTP/1.1 200 OK

{
    "state": "WA"
}

$ http https://endpoint/dev/cities/portland
HTTP/1.1 200 OK

{
    "state": "OR"
}

Notice what happens if we try to request a city that’s not in our CITIES_TO_STATE map:

$ http https://endpoint/dev/cities/vancouver
HTTP/1.1 500 Internal Server Error
Content-Type: application/json
X-Cache: Error from cloudfront

{
    "Code": "ChaliceViewError",
    "Message": "ChaliceViewError: An internal server error occurred."
}

In the next section, we’ll see how to fix this and provide better error messages.

Tutorial: Error Messages

In the example above, you’ll notice that when our app raised an uncaught exception, a 500 internal server error was returned.

In this section, we’re going to show how you can debug and improve these error messages.

The first thing we’re going to look at is how we can debug this issue. By default, debugging is turned off, but you can enable debugging to get more information:

from chalice import Chalice

app = Chalice(app_name='helloworld')
app.debug = True

The app.debug = True enables debugging for your app. Save this file and redeploy your changes:

$ chalice deploy
...
https://endpoint/dev/

Now, when you request the same URL that returned an internal server error, you’ll get back the original stack trace:

$ http https://endpoint/dev/cities/vancouver
Traceback (most recent call last):
  File "/var/task/chalice/app.py", line 304, in _get_view_function_response
    response = view_function(*function_args)
  File "/var/task/app.py", line 18, in state_of_city
    return {'state': CITIES_TO_STATE[city]}
KeyError: u'vancouver'

We can see that the error is caused from an uncaught KeyError resulting from trying to access the vancouver key.

Now that we know the error, we can fix our code. What we’d like to do is catch this exception and instead return a more helpful error message to the user. Here’s the updated code:

from chalice import BadRequestError

@app.route('/cities/{city}')
def state_of_city(city):
    try:
        return {'state': CITIES_TO_STATE[city]}
    except KeyError:
        raise BadRequestError("Unknown city '%s', valid choices are: %s" % (
            city, ', '.join(CITIES_TO_STATE.keys())))

Save and deploy these changes:

$ chalice deploy
$ http https://endpoint/dev/cities/vancouver
HTTP/1.1 400 Bad Request

{
    "Code": "BadRequestError",
    "Message": "Unknown city 'vancouver', valid choices are: portland, seattle"
}

We can see now that we have received a Code and Message key, with the message being the value we passed to BadRequestError. Whenever you raise a BadRequestError from your view function, the framework will return an HTTP status code of 400 along with a JSON body with a Code and Message. There are a few additional exceptions you can raise from your python code:

* BadRequestError - return a status code of 400
* UnauthorizedError - return a status code of 401
* ForbiddenError - return a status code of 403
* NotFoundError - return a status code of 404
* ConflictError - return a status code of 409
* TooManyRequestsError - return a status code of 429
* ChaliceViewError - return a status code of 500

You can import these directly from the chalice package:

from chalice import UnauthorizedError

Tutorial: Additional Routing

So far, our examples have only allowed GET requests. It’s actually possible to support additional HTTP methods. Here’s an example of a view function that supports PUT:

@app.route('/resource/{value}', methods=['PUT'])
def put_test(value):
    return {"value": value}

We can test this method using the http command:

$ http PUT https://endpoint/dev/resource/foo
HTTP/1.1 200 OK

{
    "value": "foo"
}

Note that the methods kwarg accepts a list of methods. Your view function will be called when any of the HTTP methods you specify are used for the specified resource. For example:

@app.route('/myview', methods=['POST', 'PUT'])
def myview():
    pass

The above view function will be called when either an HTTP POST or PUT is sent to /myview. In the next section we’ll go over how you can introspect the given request in order to differentiate between various HTTP methods.

Tutorial: Request Metadata

In the examples above, you saw how to create a view function that supports an HTTP PUT request as well as a view function that supports both POST and PUT via the same view function. However, there’s more information we might need about a given request:

• In a PUT/POST, you frequently send a request body. We need some way of accessing the contents of the request body.

• For view functions that support multiple HTTP methods, we’d like to detect which HTTP method was used so we can have different code paths for PUTs vs. POSTs.

All of this and more is handled by the current request object that the chalice library makes available to each view function when it’s called.

Let’s see an example of this. Suppose we want to create a view function that allowed you to PUT data to an object and retrieve that data via a corresponding GET. We could accomplish that with the following view function:

from chalice import NotFoundError

OBJECTS = {
}

@app.route('/objects/{key}', methods=['GET', 'PUT'])
def myobject(key):
    request = app.current_request
    if request.method == 'PUT':
        OBJECTS[key] = request.json_body
    elif request.method == 'GET':
        try:
            return {key: OBJECTS[key]}
        except KeyError:
            raise NotFoundError(key)

Save this in your app.py file and rerun chalice deploy. Now, you can make a PUT request to /objects/your-key with a request body, and retrieve the value of that body by making a subsequent GET request to the same resource. Here’s an example of its usage:

# First, trying to retrieve the key will return a 404.
$ http GET https://endpoint/dev/objects/mykey
HTTP/1.1 404 Not Found

{
    "Code": "NotFoundError",
    "Message": "mykey"
}

# Next, we'll create that key by sending a PUT request.
$ echo '{"foo": "bar"}' | http PUT https://endpoint/dev/objects/mykey
HTTP/1.1 200 OK

null

# And now we no longer get a 404, we instead get the value we previously
# put.
$ http GET https://endpoint/dev/objects/mykey
HTTP/1.1 200 OK

{
    "mykey": {
        "foo": "bar"
    }
}

You might see a problem with storing the objects in a module level OBJECTS variable. We address this in the next section.

The app.current_request object also has the following properties.

• current_request.query_params - A dict of the query params for the request.

• current_request.headers - A dict of the request headers.

• current_request.uri_params - A dict of the captured URI params.

• current_request.method - The HTTP method (as a string).

• current_request.json_body - The parsed JSON body (json.loads(raw_body))

• current_request.raw_body - The raw HTTP body as bytes.

• current_request.context - A dict of additional context information

• current_request.stage_vars - Configuration for the API Gateway stage

Don’t worry about the context and stage_vars for now. We haven’t discussed those concepts yet. The current_request object also has a to_dict method, which returns all the information about the current request as a dictionary. Let’s use this method to write a view function that returns everything it knows about the request:

@app.route('/introspect')
def introspect():
    return app.current_request.to_dict()

Save this to your app.py file and redeploy with chalice deploy. Here’s an example of hitting the /introspect URL. Note how we’re sending a query string as well as a custom X-TestHeader header:

$ http 'https://endpoint/dev/introspect?query1=value1&query2=value2' 'X-TestHeader: Foo'
HTTP/1.1 200 OK

{
    "context": {
        "apiId": "apiId",
        "httpMethod": "GET",
        "identity": {
            "accessKey": null,
            "accountId": null,
            "apiKey": null,
            "caller": null,
            "cognitoAuthenticationProvider": null,
            "cognitoAuthenticationType": null,
            "cognitoIdentityId": null,
            "cognitoIdentityPoolId": null,
            "sourceIp": "1.1.1.1",
            "userAgent": "HTTPie/0.9.3",
            "userArn": null
        },
        "requestId": "request-id",
        "resourceId": "resourceId",
        "resourcePath": "/introspect",
        "stage": "dev"
    },
    "headers": {
        "accept": "*/*",
        ...
        "x-testheader": "Foo"
    },
    "method": "GET",
    "query_params": {
        "query1": "value1",
        "query2": "value2"
    },
    "raw_body": null,
    "stage_vars": null,
    "uri_params": null
}

Tutorial: Request Content Types

The default behavior of a view function supports a request body of application/json. When a request is made with a Content-Type of application/json, the app.current_request.json_body attribute is automatically set for you. This value is the parsed JSON body.

You can also configure a view function to support other content types. You can do this by specifying the content_types paramter value to your app.route function. This parameter is a list of acceptable content types. Here’s an example of this feature:

import sys

from chalice import Chalice
if sys.version_info[0] == 3:
    # Python 3 imports.
    from urllib.parse import urlparse, parse_qs
else:
    # Python 2 imports.
    from urlparse import urlparse, parse_qs


app = Chalice(app_name='helloworld')


@app.route('/', methods=['POST'],
           content_types=['application/x-www-form-urlencoded'])
def index():
    parsed = parse_qs(app.current_request.raw_body)
    return {
        'states': parsed.get('states', [])
    }

There’s a few things worth noting in this view function. First, we’ve specified that we only accept the application/x-www-form-urlencoded content type. If we try to send a request with application/json, we’ll now get a 415 Unsupported Media Type response:

$ http POST https://endpoint/dev/ states=WA states=CA --debug
...
>>> requests.request(**{'allow_redirects': False,
 'headers': {'Accept': 'application/json',
             'Content-Type': 'application/json',
...


HTTP/1.1 415 Unsupported Media Type

{
    "message": "Unsupported Media Type"
}

If we use the --form argument, we can see the expected behavior of this view function because httpie sets the Content-Type header to application/x-www-form-urlencoded:

$ http --form POST https://endpoint/dev/formtest states=WA states=CA --debug
...
>>> requests.request(**{'allow_redirects': False,
 'headers': {'Content-Type': 'application/x-www-form-urlencoded; charset=utf-8',
...

HTTP/1.1 200 OK
{
    "states": [
        "WA",
        "CA"
    ]
}

The second thing worth noting is that app.current_request.json_body is only available for the application/json content type. In our example above, we used app.current_request.raw_body to access the raw body bytes:

parsed = parse_qs(app.current_request.raw_body)

app.current_request.json_body is set to None whenever the Content-Type is not application/json. This means that you will need to use app.current_request.raw_body and parse the request body as needed.

Tutorial: Customizing the HTTP Response

The return value from a chalice view function is serialized as JSON as the response body returned back to the caller. This makes it easy to create rest APIs that return JSON resonse bodies.

Chalice allows you to control this behavior by returning an instance of a chalice specific Response class. This behavior allows you to:

• Specify the status code to return

• Specify custom headers to add to the response

• Specify response bodies that are not application/json

Here’s an example of this:

from chalice import Chalice, Response

app = Chalice(app_name='custom-response')


@app.route('/')
def index():
    return Response(body='hello world!',
                    status_code=200,
                    headers={'Content-Type': 'text/plain'})

This will result in a plain text response body:

$ http https://endpoint/dev/
HTTP/1.1 200 OK
Content-Length: 12
Content-Type: text/plain

hello world!

Tutorial: CORS Support

You can specify whether a view supports CORS by adding the cors=True parameter to your @app.route() call. By default this value is false:

@app.route('/supports-cors', methods=['PUT'], cors=True)
def supports_cors():
    return {}

Settings cors=True has similar behavior to enabling CORS using the AWS Console. This includes:

• Injecting the Access-Control-Allow-Origin: * header to your responses, including all error responses you can return.

• Automatically adding an OPTIONS method so support preflighting requests.

The preflight request will return a response that includes:

• Access-Control-Allow-Origin: *

• The Access-Control-Allow-Methods header will return a list of all HTTP methods you’ve called out in your view function. In the example above, this will be PUT,OPTIONS.

• Access-Control-Allow-Headers: Content-Type,X-Amz-Date,Authorization, X-Api-Key,X-Amz-Security-Token.

If more fine grained control of the CORS headers is desired, set the cors parameter to an instance of CORSConfig instead of True. The CORSConfig object can be imported from from the chalice package it’s constructor takes the following keyword arguments that map to CORS headers:

Code sample defining more CORS headers:

from chalice import CORSConfig
cors_config = CORSConfig(
    allow_origin='https://foo.example.com',
    allow_headers=['X-Special-Header'],
    max_age=600,
    expose_headers=['X-Special-Header'],
    allow_credentials=True
)
@app.route('/custom_cors', methods=['GET'], cors=cors_config)
def supports_custom_cors():
    return {'cors': True}

There’s a couple of things to keep in mind when enabling cors for a view:

• An OPTIONS method for preflighting is always injected. Ensure that you don’t have OPTIONS in the methods=[...] list of your view function.

• Even though the Access-Control-Allow-Origin header can be set to a string that is a space separated list of origins, this behavior does not work on all clients that implement CORS. You should only supply a single origin to the CORSConfig object. If you need to supply multiple origins you will need to define a custom handler for it that accepts OPTIONS requests and matches the Origin header against a whitelist of origins. If the match is succssful then return just their Origin back to them in the Access-Control-Allow-Origin header.

• Every view function must explicitly enable CORS support.

The last point will change in the future. See this issue for more information.

Tutorial: Policy Generation

In the previous section we created a basic rest API that allowed you to store JSON objects by sending the JSON in the body of an HTTP PUT request to /objects/{name}. You could then retrieve objects by sending a GET request to /objects/{name}.

However, there’s a problem with the code we wrote:

OBJECTS = {
}

@app.route('/objects/{key}', methods=['GET', 'PUT'])
def myobject(key):
    request = app.current_request
    if request.method == 'PUT':
        OBJECTS[key] = request.json_body
    elif request.method == 'GET':
        try:
            return {key: OBJECTS[key]}
        except KeyError:
            raise NotFoundError(key)

We’re storing the key value pairs in a module level OBJECTS variable. We can’t rely on local storage like this persisting across requests.

A better solution would be to store this information in Amazon S3. To do this, we’re going to use boto3, the AWS SDK for Python. First, install boto3:

$ pip install boto3

Next, add boto3 to your requirements.txt file:

$ echo 'boto3==1.3.1' >> requirements.txt

The requirements.txt file should be in the same directory that contains your app.py file. Next, let’s update our view code to use boto3:

import json
import boto3
from botocore.exceptions import ClientError

from chalice import NotFoundError


S3 = boto3.client('s3', region_name='us-west-2')
BUCKET = 'your-bucket-name'


@app.route('/objects/{key}', methods=['GET', 'PUT'])
def s3objects(key):
    request = app.current_request
    if request.method == 'PUT':
        S3.put_object(Bucket=BUCKET, Key=key,
                      Body=json.dumps(request.json_body))
    elif request.method == 'GET':
        try:
            response = S3.get_object(Bucket=BUCKET, Key=key)
            return json.loads(response['Body'].read())
        except ClientError as e:
            raise NotFoundError(key)

Make sure to change BUCKET with the name of an S3 bucket you own. Redeploy your changes with chalice deploy. Now, whenever we make a PUT request to /objects/keyname, the data send will be stored in S3. Any subsequent GET requests will retrieve this data from S3.

"manage_iam_role":false
"iam_role_arn":"arn:aws:iam::<account-id>:role/<role-name>"

$ chalice gen-policy
{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Action": [
        "s3:ListAllMyBuckets"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow",
      "Sid": "9155de6ad1d74e4c8b1448255770e60c"
    }
  ]
}

$ chalice deploy
Updating IAM policy.

The following action will be added to the execution policy:

s3:ListBucket

Would you like to continue?  [Y/n]:

Tutorial: Using Custom Authentication

AWS API Gateway routes can be authenticated in multiple ways:

• API Key

• Cognito User Pools

• Custom Auth Handler

@app.route('/authenticated', methods=['GET'], api_key_required=True)
def authenticated():
    return {"secure": True}

authorizer = CognitoUserPoolAuthorizer(
    'MyPool', header='Authorization',
    provider_arns=['arn:aws:cognito:...:userpool/name'])

@app.route('/user-pools', methods=['GET'], authorizer=authorizer)
def authenticated():
    return {"secure": True}

authorizer = CustomAuthorizer(
    'MyCustomAuth', header='Authorization',
    authorizer_uri=('arn:aws:apigateway:region:lambda:path/2015-03-01'
                    '/functions/arn:aws:lambda:region:account-id:'
                    'function:FunctionName/invocations'))

@app.route('/custom-auth', methods=['GET'], authorizer=authorizer)
def authenticated():
    return {"secure": True}

Tutorial: Local Mode

As you develop your application, you may want to experiment locally before deploying your changes. You can use chalice local to spin up a local HTTP server you can use for testing.

For example, if we have the following app.py file:

from chalice import Chalice

app = Chalice(app_name='helloworld')


@app.route('/')
def index():
    return {'hello': 'world'}

We can run chalice local to test this API locally:

$ chalice local Serving on localhost:8000

We can override the port using:

$ chalice local –port=8080

We can now test our API using localhost:8000:

$ http localhost:8000/
HTTP/1.0 200 OK
Content-Length: 18
Content-Type: application/json
Date: Thu, 27 Oct 2016 20:08:43 GMT
Server: BaseHTTP/0.3 Python/2.7.11

{
    "hello": "world"
}

The chalice local command does not assume the role associated with your lambda function, so you’ll need to use an AWS_PROFILE that has sufficient permissions to your AWS resources used in your app.py.

Backlog

These are features that are in the backlog:

• Adding full support for API gateway stages - issue 20

Please share any feedback on the above issues. We’d also love to hear from you. Please create any Github issues for additional features you’d like to see: https://github.com/awslabs/chalice/issues

FAQ

Q: How does the Python Serverless Microframework for AWS compare to other similar frameworks?

The biggest difference between this framework and others is that the Python Serverless Microframework for AWS is singularly focused on using a familiar, decorator-based API to write python applications that run on Amazon API Gateway and AWS Lambda. You can think of it as Flask/Bottle for serverless APIs. Its goal is to make writing and deploying these types of applications as simple as possible specifically for Python developers.

To achieve this goal, it has to make certain tradeoffs. Python will always remain the only supported language in this framework. Not every feature of API Gateway and Lambda is exposed in the framework. It makes assumptions about how applications will be deployed, and it has restrictions on how an application can be structured. It does not address the creation and lifecycle of other AWS resources your application may need (Amazon S3 buckets, Amazon DynamoDB tables, etc.). The feature set is purposefully small.

Other full-stack frameworks offer a lot more features and configurability than what this framework has and likely will ever have. Those frameworks are excellent choices for applications that need more than what is offered by this microframework. If all you need is to create a simple rest API in Python that runs on Amazon API Gateway and AWS Lambda, consider giving the Python Serverless Microframework for AWS a try.