A limitd plug-in for hapi.js
npm i -S patova
The following code snippet shows how to register the plug-in in your server:
const Hapi = require('hapi');
const patova = require('patova');
const server = new Hapi.Server({ port: 9999 });
await server.register({
plugin: patova,
options: {
event: 'onPostAuth',
type: 'users',
limitd: limitdClient,
extractKey: (request) => request.auth.credentials
},
});
The object has the following schema (validated here using Joi):
Required
event: String
- The name of the extension point in the request lifecycle when the bucket check must be performed. Options are"onRequest"
,"onPreAuth"
,"onPostAuth"
,"onPreHandler"
(anything before the request).type: String|async (request, flowControl) => ()
- Either the bucket type as a string or a function. If you use a function, it will be called for every request, this could be a async function. It's possible to decide skip the check, for this returnflowControl.continue
;limitd
: an instance of limitd clientextractKey: (request, flowControl) => ()
- A function that receives therequest
and the objectflowControl
.request: Request
- The hapi.js request object.flowControl
- This object contains a propertycontinue
to let the plugin know that you want to skip the check. Same situation to what happen in key.
Optional
onError: (error, h) => ()
- A function that takes theerror
that occurred when trying to get a token from the bucket and thereply
interface.error: Error
- The error that occurred.h: ResponseToolkit
- The hapi.js response toolkit.
If an error occurs and no function is provided, the request lifecycle continues normally as if there was no token bucket restriction. This is a useful default behavior in case the limitd server goes down.
The following code snippet shows how to register the plug-in in your server:
const Hapi = require('hapi');
const patova = require('patova');
const server = new Hapi.Server();
server.connection({ /* options */ });
server.register({
register: patova,
options: {
event: 'onPostAuth',
type: 'users',
limitd: limitdClient,
extractKey: function(request, reply, done){
var key = request.auth.credentials.userId;
done(null, key);
}
},
}, err => {
//
});
The object has the following schema (validated here using Joi):
Required
event: String
- The name of the extension point in the request lifecycle when the bucket check must be performed. Options are"onRequest"
,"onPreAuth"
,"onPostAuth"
,"onPreHandler"
(anything before the request).type: String|(request, callback) => ()
- Either the bucket type as a string or a function. If you use a function, it will be called for every request, this function must invoke the callback function when it is finished.limitd
: an instance of limitd clientextractKey: (request, reply, done) => ()
- A function that receives therequest
and a callbackdone
.request: Request
- The hapi.js request object.reply: Reply
- The hapi.js reply interface. Useful if you want to skip the check.done: (err: Error, key: String)
- A function that takes an error as the first parameter and the bucket key as the second parameter.
Optional
onError: (error, reply) => ()
- A function that takes theerror
that occurred when trying to get a token from the bucket and thereply
interface.error: Error
- The error that occurred.reply: Reply
- The hapi.js reply interface.
If an error occurs and no function is provided, the request lifecycle continues normally as if there was no token bucket restriction. This is a useful default behavior in case the limitd server goes down.
Feel free to open issues with questions/bugs/features. PRs are also welcome.
Hapi Version | Patova Version |
---|---|
Hapi <= 16 | v2.2.x |
Hapi >= 17 | v3.0.x |
MIT