reCAPTCHA Fastly Compute@Edge Library

A library to access reCAPTCHA Enterprise via Fastly Compute@Edge.
Usage
This project is intended to be used in one of two ways:
- Using a prebuilt package uploaded to your Fastly project.
- or imported as an NPM package for advanced use-cases.
Prebuilt Package
Check the Releases page for the most recent build for Fastly.
This package is intended to be used in concert with reCAPTCHA Firewall Policies.
Each Firewall Policy rule has a path, a condition and a set of actions.
- Paths are written as glob patterns matching an incoming request.
- Examples: "/login.html", "/pages/*.php", "/static/**/*"
- Conditions are written using CEL expression language with variables populated from the incoming Request and reCAPTCHA evaluation.
- Examples: "recaptcha.score >= 0.7", "!recaptcha.token.valid", "http.ip.startsWith("192.168.0")"
- Actions may be Allow, Block, Show a reCAPTCHA Challenge Page, Substitute a different page, Set a Request Header on the Request to the origin backend.
The following actions are valid:
- For an 'Allow' action, the request will continue to the origin backend.
- This is used for normal traffic.
- For a 'Block' action, a 403 will be returned to the user. The origin is never called.
- This may be used to block expected bot traffic.
- For a 'Redirect' action, a synthetic page with a reCAPTCHA challenge will be returned. The origin is never called.
- This may be used to add friction to expected bot traffic, or gain further confidence in human traffic.
- For a 'SetHeader' action, a request header will be added to the origin request. This can be used to communicate information to the backend or trigger application specific protections.
The policy logic to decide the action to take is approximately as demonstrated below:
def decideActions(request, policies):
for policy in policies:
if policy.path.matches(request.path) and
policy.condition.evaluateOn(request) == True:
return policy.actions
return [new AllowAction()]
When deployed as a Fastly Compute service, this package will cache a list of your configured Firewall Policies. When an incoming request is received, first the requested path
will be checked against the list of all policies.
If no policy paths match, reCAPTCHA will be bypassed and the request will be forwarded to the origin (the Allow action).
If any policy paths match the incoming request,
the reCAPTCHA CreateAssessment API will be called. reCAPTCHA will evaluate the policies as per the above logic and a list of actions will be returned to the Fastly Compute service. The Compute service
will execute the actions.
To integrate this package with an existing Fastly account:
- Create the appropriate reCAPTCHA Site Keys in Google Cloud reCAPTCHA Console.
- Upload the package to Fastly with the
fastly compute deploy
CLI command or in the 'Package' section of the new Compute service in the Fastly web UI.
- In the Fastly web UI, add the relevant site keys and variables in the "Dictionaries" section of the new worker.
- Create a set of Firewall Policies to protect sensative pages or actions.
- Update your DNS entries to direct traffic to the new service.
Please see the reCAPTCHA Google Cloud Documentation for more details on each step.
As a Library
This package has been added to the NPM package repo as @google-cloud/recaptcha-fastly.
This library supports a standard reCAPTCHA v2 or v3 workflow, and is intended to be used on the Fastly Compute edge compute platform. To use this library, first create a FastlyContext
object. This object
can be initialized with a Fastly ConfigStore or
inline constants:
import {
FastlyContext,
recaptchaConfigFromConfigStore,
createAssessment
} from "@google-cloud/recaptcha-fastly";
addEventListener("fetch", (event) => event.respondWith(handleRequest(event)));
async function handleRequest(event: FetchEvent): Promise<Response> {
const rcctx = new FastlyContext(event, recaptchaConfigFromConfigStore("recaptcha"));
// OR: initialized inline
const rcctx = new FastlyContext(env, ctx, {projectNumber: 12345, apiKey: "abcd", enterpriseSiteKey: "6Labcdefg"});
... // do further processing
}
This context can then be used to call reCAPTCHA's CreateAssessment
on the incoming Request:
...
const assessment = await createAssessment(rcctx, request);
...
Most common request data expected in CreateAssessment
will be automatically populated when calling the createAssessment
function, including:
- userAgent
- userIpAddress
- requestedUri
- ja3 or ja4
- headers
The Project number and API Key set when creating the FastlyContext
will be used to form the correct CreateAssessment
endpoint URL. The following URL format will be used: https://recaptchaenterprise.googleapis.com/v1/projects/{projectNumber}/assessments??key={apikey}
.
The enterpriseSiteKey
set when creating the FastlyContext
will be used to populate the Assessment event.
The user's reCAPTCHA token will be automatically extracted from the incoming request body if all of these conditions are true:
- The incoming requests uses a POST HTTP method
- The content type is
application/json
or application/x-www-form-urlencoded
or multipart/form-data
- The token is expected to reside in the
g-recaptcha-response
field.
If all of these cases are true, simply call the createAssessment method as above.
If one or more of the token format cases are false, the token must be manually extracted and passed as an additional parameter:
...
const token = manuallyExtractToken(request); // You must define this function.
const assessment = await createAssessment(rcctx, request, {token});
...
The expectedAction
parameter is not automatically populated, and should be populated if applicable.
const assessment = await createAssessment(rcctx, request, {expectedAction: "login"});
See the official documentation on action names.
It is important that createAssessment is only called on paths where you expect a user to pass a token.
import {
FastlyContext,
createAssessment,
pathMatch,
} from "@google-cloud/recaptcha-fastly";
...
if (pathMatch(request, "/login", "POST")) {
const assessment = await createAssessment(rcctx, request, {expectedAction: "login"});
... // check assessment results, such as score.
}
...
This will avoid the added latency from the CreateAssessment RPC, and reduce billing events.
For complete end-to-end examples, see the examples directory.
Contribution
Please see our Contribution guidelines.
Issues and Support
For technical issues, please see the reCAPTCHA Enterprise Support Documentation.
For bugs or issues specifically with this codebase, please open a new Github issue in this project.