happn-3
v13.12.1
Published
pub/sub api as a service using primus and mongo & redis or nedb, can work as cluster, single process or embedded using nedb
Downloads
3,956
Readme
Introduction
Happn is a mini database combined with pub/sub, the system stores json objects on paths. Paths can be queried using wildcard syntax. The happn client can run in the browser or in a node process. Happn clients can subscribe to events on paths, events happn when data is changed by a client on a path, either by a set or a remove operation.
Happn stores its data in a collection called 'happn' by default on your mongodb/nedb. The happn system is actually built to be a module, this is because the idea is that you will be able to initialize a server in your own code, and possibly attach your own plugins to various system events.
A paid for alternative to happn would be firebase
Getting started
npm install happn-3
You need NodeJS and NPM of course, you also need to know how node works (as my setup instructions are pretty minimal) To run the tests, clone the repo, npm install then npm test:
git clone https://github.com/happner/happn-3.git
npm install
npm test
But if you want to run your own service do the following: Create a directory you want to run your happn in, create a node application in it - with some kind of main.js and a package.json
In node_modules/happn/test in your folder, the 1_eventemitter_embedded_sanity.js and 2_websockets_embedded_sanity.js scripts demonstrate the server and client interactions shown in the following code snippets
configuration
for full service and client configuration options
starting service:
The service runs on port 55000 by default - the following code snippet demonstrates how to instantiate a server.
var happn = require('happn-3')
var happnInstance; //this will be your server instance
//we are using a compact default config here, port defaults to 55000
happn.service.create({
utils: {
logLevel: 'error',
// see happn-logger module for more config options
}
},
function (e, happn) {
if (e)
return callback(e);
happnInstance = happn; //here it is, your server instance
happnInstance.log.info('server up');
});
In your console, go to your application folder and runnode mainyour server should start up and be listening on your port of choice.
THE HAPPN CLIENT
Using node:
var happn = require('happn-3');
var myClient; //this will be your client instance
/**
example options are :
{
host: "127.0.0.1", //(default)
port: 55000, //(default)
username: 'username', //only necessary if server is secure
password: 'password', //only necessary if server is secure
socket: {
pingTimeout: 45e3 // 45 seconds by default, if set to false the client
// will not detect connection failures and emit the
// 'reconnect scheduled event'
},
// on socket error:
reconnectWait: 1e3, // wait 1 seconds before reconnecting on socket error
retryOnSocketErrorMaxInterval: 120e3, // maximum interval between reconnects when there is a socket error
}
**/
const myClientInstance = await HappnClient.create(options);
To use the browser client, make sure the server is running, and reference the client javascript with the url pointing to the running server instances port and ip address like so:
<script type="text/javascript" src="http://localhost:55000/browser_client"></script>
<script>
const myClientInstance = await HappnClient.create(options);
</script>
Intra-process / local client:
although we have direct access to the services (security included) - this method still requires a username and password if the happn instance is secure
service.create(function (e, happnInst) {
if (e) throw e;
happnInstance = happnInst;
happnInstance.services.session.localClient(/*credentials argument only necessary if secure*/{username:'test', password:'test'}, function(e, instance){
var myLocalClient = instance;
//myLocalClient.set(...)
});
});
Intra-process / local admin client:
will pass back a client with admin rights no username or password necessary, this is because we have direct access to the services, security included
service.create(function (e, happnInst) {
if (e) throw e;
happnInstance = happnInst;
happnInstance.services.session.localAdminClient(function(e, instance){
var myLocalAdminClient = instance;
//myLocalClient.set(...)
});
});
##NB: NODE_ENV environment variable Set your NODE_ENV variable to "production" when running happn in a production environment, otherwise your browser client file will be regenerated every time the happn server is restarted.
SET
Puts the json in the branch e2e_test1/testsubscribe/data, creates the branch if it does not exist
//the noPublish parameter means this data change wont be published to other subscribers, it is false by default
//there are a bunch other parameters - like noStore (the json isnt persisted, but the message is published)
myClient.set('e2e_test1/testsubscribe/data/', {property1:'property1',property2:'property2',property3:'property3'}, {noPublish:true}, function(e, result){
//your result object has a special _meta property that contains its actual _id, path, created and modified dates
//so you get back {property1:'property1',property2:'property2',property3:'property3', _meta:{path:'e2e_test1/testsubscribe/data/', created:20151011893020}}
});
NB - by setting the option merge:true, the data at the end of the path is not overwritten by your json, it is rather merged with the data in your json, overwriting the fields you specify in your set data, but leaving the fields that are already at that branch.
PUBLISH
publishes the json to all topic subscribers that match e2e_test1/testsubscribe/data, the data is not stored or returned in the response, only the _meta is returned
myClient.publish('e2e_test1/testsubscribe/data/', {property1:'property1',property2:'property2',property3:'property3'}, function(e, result){
//your result does not contain the changed data, but it still has the _meta property:
result = {
_meta:{
published: true,
type: 'response',
status: 'ok',
eventId: 4, //eventId matching event handler on client
sessionId: '[guid: your current session id]'
}
}
});
INCREMENT
allows a counter to be incremented by an increment value
single guage - causing on event:
//listen on a path
myClient.on('my/increment/guage', function(data){
//NB; the data on the event will look like this
//{guage:'counter', value:1}
myClient.get('my/increment/guage', function(e, gotIncrementedData){
expect(gotIncrementedData[data.value].value).to.be(1);
});
}, function(e){
if (e) throw e;
//increment convenience method
myClient.increment('my/increment/guage', 1, function(e){
if (e) throw e;
});
});
increment multiple times, guage defaults to counter and increment value is 1:
var async = require('async');
async.timesSeries(10, function (time, timeCB) {
myClient.increment('my/guage', function (e) {
timeCB(e);
});
}, function (e) {
myClient.get('my/guage', function (e, result) {
expect(result['counter-0'].value).to.be(10);
});
});
multiple guages on the same path:
var async = require('async');
async.timesSeries(10, function (time, timeCB) {
myClient.increment('my/dashboard', 'counter-' + time, 1, function (e) {
timeCB(e);
});
}, function (e) {
myClient.get('my/dashboard', function (e, result) {
expect(result['counter-0'].value).to.be(1);
expect(result['counter-1'].value).to.be(1);
expect(result['counter-2'].value).to.be(1);
expect(result['counter-3'].value).to.be(1);
expect(result['counter-4'].value).to.be(1);
expect(result['counter-5'].value).to.be(1);
expect(result['counter-6'].value).to.be(1);
expect(result['counter-7'].value).to.be(1);
expect(result['counter-8'].value).to.be(1);
expect(result['counter-9'].value).to.be(1);
});
});
decrement a guage with a minus value:
var incrementCount = 0;
//listening on the event
myClient.on('my/test/guage', function (data) {
incrementCount++;
if (incrementCount == 1){
expect(data.value).to.be(3);
expect(data.guage).to.be('custom');
}
if (incrementCount == 2){
expect(data.value).to.be(1);
expect(data.guage).to.be('custom');
}
}, function (e) {
myClient.increment('my/test/guage', 'custom', 3, function (e) {
myClient.get('my/test/guage', function (e, result) {
expect(result['custom'].value).to.be(1);
myClient.increment('my/test/guage', 'custom', -2, function (e) {
myClient.get('my/dashboard', function (e, result) {
expect(result['custom'].value).to.be(1);
});
});
});
});
});
GET
Gets the data living at the specified branch
myClient.get('e2e_test1/testsubscribe/data',
null, //options
function(e, results){
//results is your data, if you used a wildcard in your path, you get back an array
//if you used an explicit path, you get back your data as the object on that path
You can also use wildcards, gets all items with the path starting e2e_test1/testsubscribe/data
myClient.get('e2e_test1/testsubscribe/data*',
null,
function(e, results){
//results is your data
results.map(function(item){
});
You can also just get paths, without data
myClient.getPaths('e2e_test1/testwildcard/*', function(e, results){
SEARCH
You can pass mongo style search parameters to look for data sets within specific key ranges, using limit and skip
var options = {
fields: {"name": 1},
sort: {"name": 1},
limit: 10,
skip:5
}
var criteria = {
$or: [{"region": {$in: ['North', 'South', 'East', 'West']}},
{"town": {$in: ['North.Cape Town', 'South.East London']}}],
"surname": {$in: ["Bishop", "Emslie"]}
}
publisherclient.get('/users/*', {
criteria: criteria,
options: options
},
function (e, search_results) {
//and your results are here
search_results.map(function(user){
if (user.name == 'simon')
throw new Error('stay away from this chap, he is dodgy');
});
}
);
Using $regex is a bit different, you can pass in a regex string representation or an array with regex string as the first element and modifiers such as i as the second element depending on whether you want to pass in regex arguments such as case insensitive (i)
var options = {
fields: {"name": 1},
sort: {"name": 1},
limit: 1
}
var criteria = {
"name": {
"$regex": [".*simon.*", "i"]//array with regex first, then "i" argument, does Regex.apply(null, [your array]) in backend
}
};
//alternatively - for a case sensitive simpler search:
var criteriaSimple = {
"name": {
"$regex": ".*simon.*"//array with regex first, then "i" argument, does Regex.apply(null, [your array]) in backend
}
};
publisherclient.get('/users/*', {
criteria: criteria,//(or criteriaSimple)
options: options
},
function (e, search_results) {
//and your results are here
search_results.map(function(user){
if (user.name == 'simon')
throw new Error('stay away from this chap, he is dodgy');
});
}
);
You can also use skip and limit for paging:
var totalRecords = 100;
var pageSize = 10;
var expectedPages = totalRecords / pageSize;
var indexes = [];
for (let i = 0; i < totalRecords; i++) indexes.push(i);
for (let index of indexes){
await searchClient.set('series/horror/' + index, {
name: 'nightmare on elm street',
genre: 'horror',
episode:index
});
await searchClient.set('series/fantasy/' + index, {
name: 'game of thrones',
genre: 'fantasy',
episode:index
});
}
var options = {
sort: {
"_meta.created": -1
},
limit: pageSize
};
var criteria = {
"genre": "horror"
};
var foundPages = [];
for (let i = 0; i < expectedPages; i++){
options.skip = foundPages.length;
let results = await searchClient.get('series/*', {
criteria: criteria,
options: options
});
foundPages = foundPages.concat(results);
}
let allResults = await searchClient.get('series/*', {
criteria: criteria,
options: {
sort: {
"_meta.created": -1
}
}
});
expect(allResults.length).to.eql(foundPages.length);
expect(allResults).to.eql(foundPages);
DELETE / REMOVE
deletes the data living at the specified branch
myClient.remove('/e2e_test1/testsubscribe/data/delete_me', null, function(e, result){
if (!e)
//your item was deleted, result.payload is an object that lists the amount of objects deleted
EVENTS
you can listen to any SET & REMOVE events happening in your data - you can specify a path you want to listen on or you can listen to all SET and DELETE events using a catch-all listener, the * character denotes a wildcard
NB about wildcards:
As of version 8.0.0 the wildcard is a whole word, and the / is used to denote path segments - ie: to get all events for a set or remove with path /my/test/event you need to subscribe to /my/*/*, /my/* and /my* or /my/te*/event will no longer work. One deviation from this limitation is in the usage of variable depth subscriptions (as documented below).
Specific listener:
myClient.on('/e2e_test1/testsubscribe/data/delete_me', //the path you are listening on
{event_type:'remove', // either set, remove or all - defaults to all
count:0},// how many times you want your handler to handle for before it is removed - default is 0 (infinity)
function(//your listener event handler
message, //the actual object data being set or removed
meta){ //the meta data - path, modified,created _id etc.
},
function(e){
//passes in an error if you were unable to register your listener
});
//this is now promise based as of v11.5.0
const handle = await myClient.on('/e2e_test1/testsubscribe/data/delete_me', //the path you are listening on
{event_type:'remove', // either set, remove or all - defaults to all
count:0},// how many times you want your handler to handle for before it is removed - default is 0 (infinity)
function(//your listener event handler
message, //the actual object data being set or removed
meta){ //the meta data - path, modified,created _id etc.
//event happened
});
await myClient.off(handle); //unsubscribe
Catch all listener:
myClient.onAll(function(//your listener event handler
message, //the actual object data being set or removed
meta){
//the meta data - path, modified,created _id, also tells you what type of operation happened - ie. GET, SET etc.
},
function(e){
//passes in an error if you were unable to register your listener
});
//this is now promise based as of v11.5.0
const handle = await myClient.onAll(function(//your listener event handler
message, //the actual object data being set or removed
meta){ //the meta data - path, modified,created _id, also tells you what type of operation happened - ie. GET, SET etc.
});
Once listener:
const handle = await myClient.once('/e2e_test1/testsubscribe/data/delete_me', //the path you are listening on
{ event_type:'*' }, // either set, remove or all - defaults to all
function(//your listener event handler
message, //the actual object data being set or removed
meta){ //the meta data - path, modified,created _id etc.
//event happened
});
await myClient.off(handle); //unsubscribe, these will auto-expire after they have received a single message
EVENT DATA
you can grab the data you are listening for immediately either by causing the events to be emitted immediately on successful subscription or you can have the data returned as part of the subscription callback using the initialCallback and initialEmit options respectively
//get the data back as part of the subscription callback
listenerclient.on('/e2e_test1/testsubscribe/data/values_on_callback_test/*',
{"event_type": "set",
"initialCallback":true //set to true, causes data to be passed back
}, function (message) {
expect(message.updated).to.be(true);
callback();
}, function(e, reference, response){
if (e) return callback(e);
try{
//the response is your data, ordered by modified - will always be in an array even if only one or none is found
expect(response.length).to.be(2);
expect(response[0].test).to.be('data');
expect(response[1].test).to.be('data1');
listenerclient.set('/e2e_test1/testsubscribe/data/values_on_callback_test/1', {"test":"data", "updated":true}, function(e){
if (e) return callback(e);
});
}catch(e){
return callback(e);
}
});
//get the data emitted back immediately
listenerclient.on('/e2e_test1/testsubscribe/data/values_emitted_test/*',
{"event_type": "set",
"initialEmit":true //set to true causes emit to happen on successful subscription
}, function (message, meta) {
//this emit handler runs immediately
caughtEmitted++;
if (caughtEmitted == 2){
expect(message.test).to.be("data1");
callback();
}
}, function(e){
if (e) return callback(e);
});
UNSUBSCRIBING FROM EVENTS
//use the .off method to unsubscribe from a specific event (the handle is returned by the .on callback) or the .offPath method to unsubscribe from all listeners on a path:
var currentListenerId;
var onRan = false;
var pathOnRan = false;
listenerclient.on('/e2e_test1/testsubscribe/data/on_off_test', {event_type: 'set', count: 0}, function (message) {
if (pathOnRan) return callback(new Error('subscription was not removed by path'));
else pathOnRan = true;
//NB - unsubscribing by path
listenerclient.offPath('/e2e_test1/testsubscribe/data/on_off_test', function (e) {
if (e)
return callback(new Error(e));
listenerclient.on('/e2e_test1/testsubscribe/data/on_off_test', {event_type: 'set', count: 0},
function (message) {
if (onRan) return callback(new Error('subscription was not removed'));
else {
onRan = true;
//NB - unsubscribing by listener handle
listenerclient.off(currentListenerId, function (e) {
if (e)
return callback(new Error(e));
publisherclient.set('/e2e_test1/testsubscribe/data/on_off_test', {"test":"data"}, function (e, setresult) {
if (e) return callback(new Error(e));
setTimeout(callback, 2000);
});
});
}
},
function (e, listenerId) {
//NB - listener id is passed in on the .on callback
if (e) return callback(new Error(e));
currentListenerId = listenerId;
publisherclient.set('/e2e_test1/testsubscribe/data/on_off_test', {"test":"data"}, function (e, setresult) {
if (e) return callback(new Error(e));
});
});
});
}, function (e, listenerId) {
if (e) return callback(new Error(e));
currentListenerId = listenerId;
publisherclient.set('/e2e_test1/testsubscribe/data/on_off_test', {"test":"data"}, function (e) {
if (e) return callback(new Error(e));
});
});
TARGETING EVENTS
sets and removes can be targeted for a specific client session, if you have access to a client session id, or need to do a return-ticket post, you can add the session id's you want your event data to go to to the targetClients option
var mySessionId = myClient.sesson.id;
//only publish to myself:
other_client_instance.on('for/my/eyes/only', function(data){
//should NOT receive this
});
myClient.on('for/my/eyes/only', function(data){
//should receive this
});
myClient.set('for/my/eyes/only', {property1:'property1'}, {targetClients:[mySessionId]}, function(e, result){
...
});
EVENTS WITH CUSTOM META
sets and removes can declare custom metadata that will be sent to subscribers
client.set('/some/topic', {DATA: 1}, {meta: {custom: 1}}, function(e) {})
// elsewhere
client.on('/some/topic', function(data, meta) {
meta.custom == 1;
});
Reserved meta key names will have no effect. ('created','modified','path','action','type','published','status','eventId','sessionId')
MERGE SUBSCRIPTIONS
you can subscribe to data changes on set/remove and specify only to recieve the merge data as posted in the set operation with the merge:true option, NB: this is not a true delta, as you may receive some duplicate input fields
myClient.on('/merge/only/path', {
event_type: 'set',
merge: true
}, function (message) {
console.log('emit happened - message is {some:"data"}');
}, function (e) {
console.log('subscription happened');
});
myClient.set('/merge/only/path',
{some:"data"},
{merge:true},
function (e) {
console.log('set happened');
});
VARIABLE DEPTH SUBSCRIPTIONS
A special subscription, with a trailing /** on the path, allows for subscriptions to multiple wildcard paths, up to a specific depth
const happn = require('happn-3');
//NB the default variable depth is 5, you can set it when initialising the client like so:
const myClient = await happn.client.create({config:{username:'_ADMIN', password:'happn', defaultVariableDepth:10}});
var handler = function(data){
};
myClient.on('/test/path/**', { depth:4 }, handler, function(e, variableDepthHandle){
//you can unsubscribe as per normal
// ie: myClient.off(variableDepthHandle)
});
//is the same as
myClient.on('/test/path/*', handler, function(e){
});
myClient.on('/test/path/*/*', handler, function(e){
});
myClient.on('/test/path/*/*/*', handler, function(e){
});
myClient.on('/test/path/*/*/*/*', handler, function(e){
});
//NB: up to a depth of 4, so the event will not fire for a larger depth, ie: /test/path/1/2/3/4/5
//NB: this functionality also works with initialCallback and initialEmit
myClient.on('/test/path/**', {
"event_type": "set",
"initialEmit": true
}, function (message, meta) {
//items will be immediately emitted up to the depth specified
});
myClient.on('/test/path/**', {
"event_type": "set",
"initialCallback": true
}, function (message) {
expect(message.updated).to.be(true);
callback();
}, function (e, reference, response) {
//response will be an array of items that exist to the specified depth
});
MERGING
you can do a set command and specify that you want to merge the json you are pushing with the existing dataset, this means any existing values that are not in the set json but exist in the database are persisted
myClient.set('e2e_test1/testsubscribe/data/', {property1:'property1',property2:'property2',property3:'property3'}, {merge:true}, function(e, result){
});
SESSION AND CONNECTION EVENTS:
these events are emitted if the client connection state or session state changes
// session-ended event is emitted if the session is disconnected from the server side
myClient.onEvent('session-ended', (evt) => {
//evt.reason could be:
//inactivity-threshold - the client has been inactive for a period exceeding what the session is profiled for (see profiles)
//session-revoked - the client session has been revoked on the server side
//security directory update: user deleted - the user that the session is associated with has been deleted on the server
});
// reconnect-scheduled event is emitted if the connection with the server has been interrupted
myClient.onEvent('reconnect-scheduled', (evt) => {
});
// reconnect-successful event is emitted if the connection with the server has been restored
myClient.onEvent('reconnect-successful', (evt) => {
});
After version 11.6.0, by default some basic session info is logged whenever a client attached or detaches in the format, as stringified JSON:
{
"event", //session attached / session detatched
"username", //user name or 'anonymous (unsecure connection)',
"sourceAddress", //session source address,
"sourcePort", //client side port
"upgradeUrl",// primus upgrade url for establishing the connection socket
"happnVersion", // for new clients, package version
"happnProtocolVersion" // happn_4, happn (old connections)
}
This can be switched off by updating the session service config:
const Happn = require('happn-3');
let myService = await Happn.service.create({
name: 'TEST-NAME',
secure: true,
services: {
session: {
config: {
disableSessionEventLogging:true
}
}
}
});
CLIENT DISCONNECTION AND LOGOUT
the happn client can disconnect using a plain disconnect() method (Promise) - or you can disconnect with options, one of which is revokeToken:true, this will disconnect the client and revoke the access token for the client - all other clients created with the same token will be disconnected as well, there is a convenience logout() method that is in effect the same as disconnect({ revokeToken: true })
const happn = require('happn-3');
const myClient = await happn.client.create({ username:'_ADMIN', password:'happn'});
const myOtherClient = await happn.client.create({ username:'_ADMIN', token: myClient.session.token });
// plain disconnect
await myClient.disconnect();
// the following methods will revoke the token and cause myOtherClient to be automatically disconnected:
// disconnect with revokeToken:true option
await myClient.disconnect({ revokeToken: true });
// logout convenience method, also revokes the token
await myClient.logout();
INFINITE TOKEN REVOCATION
revoking a token that is configured with an infinite ttl or ttl with a value of 0 comes with the risk of having an ever growing token revocation list, this could cause the system to eventually run out of memory
// by default the system will allow revocation of infinite ttl tokens
// to configure the security service to disallow, set the property allowTTL0Revocations to false for the security service config:
const happn = require('happn-3');
const happnInstance = await happn.service.create({
secure: true,
services: {
security: {
config: {
allowTTL0Revocations: false, // the service will throw and bubble up an error if the client attempts to revoke an infinite ttl token
profiles: [
//profiles are in an array, in descending order of priority, so if you fit more than one profile, the top profile is chosen
{
name: 'infinite-token', // our intentionally long lived profile (assuming this user exists)
session: {
'user.username': 'LONG_LIVED_USER',
},
policy: {
ttl: Infinity,
},
},
],
},
},
},
});
const credentials = { username: 'LONG_LIVED_USER', password: 'xxx' };
await happnInstance.services.security.users.upsertUser(credentials);
const myClient = await happn.client.create(credentials);
try {
// will throw an error, as the server does not allow ttl 0 tokens to be revoked:
await myClient.logout();
console.log('UNEXPECTED SUCCESS');
} catch (e) {
// e.message should be 'revoking a token without a ttl means it stays in the revocation list forever'
console.log(`EXPECTED FAILURE: ${e.message}`);
await happnInstance.stop();
}
SECURITY
happn server instances can be secured with user and group authentication and authorisation, a default user and group called _ADMIN is created per happn instance, the admin password is 'happn' but is configurable (MAKE SURE PRODUCTION INSTANCES DO NOT RUN OFF THE DEFAULT PASSWORD)
var happn = require('happn-3');
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, instance) {
if (e)
return callback(e);
happnInstance = instance; //here it is, your server instance
});
at the moment, adding users, groups and permissions can only be done by directly accessing the security service, to see how this is done - please look at the functional test for security
SECURITY CLIENT
the client needs to be instantiated with user credentials and with the secure option set to true to connect to a secure server
//logging in with the _ADMIN user
var happn = require('happn-3');
happn.client.create({username:'_ADMIN', password:'testPWD', secure:true},function(e, instance) {
SECURITY USERS AND GROUPS
to modify users and groups, a direct code based connection to the happn-3 security service is required, thus users and groups should not be modified in any way over the wire
add a group
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
var myGroup = {
name:'TEST',
permissions:{
'/test/path/*':{actions:['get', 'set']}, //allow only gets and sets to this path
'/test/allow/all':{actions:['*']} //allow all actions to this path
},
custom_data:{//any custom data you want
test:'data'
}
};
//NB! permissions are stored separately to the group, so when upserting the group and it allready exists
//with other permissions the current upserts permissions are merged with the existing ones, down to action level
myHappn3Instance.services.security.groups.upsertGroup(myGroup)
.then(function(upserted){
//group added
})
});
list groups
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
var myGroup = {
name:'TEST',
permissions:{
'/test/path/*':{actions:['get', 'set']}, //allow only gets and sets to this path
'/test/allow/all':{actions:['*']} //allow all actions to this path
},
custom_data:{//any custom data you want
test:'data'
}
};
//NB! permissions are stored separately to the group, so when upserting the group and it allready exists
//with other permissions the current upserts permissions are merged with the existing ones, down to action level
myHappn3Instance.services.security.groups.upsertGroup(myGroup)
.then(function(upserted){
//group added
return myHappn3Instance.services.security.groups.listGroups('TES*', {
criteria:{
name:{$eq:'TEST'}
},
/* optional
skip:2,
limit:5,
count:true //will only return the groups count
*/
});
})
.then(function(group){
//group just added would be returned
});
});
if we are using mongodb, we are able to specify collation for listing users see the mongo tests.
NB! permissions are separate to the group, so when upserting the group and it already exists with other permissions the current upserts permissions are merged with the existing ones, down to action level
add a user
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
var myUser = {
username: 'TEST',
password: 'TEST PWD',
custom_data: {
something: 'usefull'
}
};
myHappn3Instance.services.security.users.upsertUser(myUser)
.then(function(upserted){
//user added, with no permissions yet - permissions must be assigned to the user by linking the user to a group
})
});
link a fetched user to a fetched group
demonstrates getUser, getGroup and linkGroup
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
var myUser, myGroup;
myHappn3Instance.services.security.users.getUser('TEST')
.then(function(user){
myUser = user;
return myHappn3Instance.services.security.groups.getGroup('TEST');
})
.then(function(group){
myGroup = group;
return myHappn3Instance.services.security.groups.linkGroup(myGroup, myUser);
})
.then(function(){
//your TEST user now has the permissions assigned by your TEST group
});
});
list users
user can be listed by group name (exact match) or by a username (partial match with wildcard - with optional additional criteria)
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
//by username, with more specific criteria (mongo style)
myHappn3Instance.services.security.users.listUsers('TEST*', {
criteria:{username:{$eq:'TESTUSER1'}},
/* optional
skip:2,
limit:5,
count:true //will only return the users count
*/
})
.then(function(users){
//returns:
// [
// {username:'test1', custom_data:{test:1}},
// {username:'test2', custom_data:{test:1}},
// {username:'test3', custom_data:{test:1}}
// ]
//by group name, note optional criteria
return myHappn3Instance.services.security.users.listUsersByGroup('TEST', {criteria:{custom_data:1}})
})
.then(function(users){
//returns:
// [
// {username:'test1', custom_data:{test:1}}
// ]
//much faster - just get the usernames by the group name
return myHappn3Instance.services.security.users.listUserNamesByGroup('TEST');
})
.then(function(usernames){
//returns:
// [
// 'test1'
// ]
});
});
upsert a permission
permissions can be merged by saving a group, or permissions can be added to a group piecemeal in the following way:
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
myHappn3Instance.services.security.groups.upsertPermission('TEST'/* group name*/, '/test/path/*' /* permission path */, 'on' /* action */, true /* allow (default) - if false the permission is removed */)
.then(function () {
//users belonging to the TEST group can now do "set", "get" AND "on" operations as opposed to only set and get (check above addGroup example to double check the pre-existing permissions)
});
});
remove a permission
permissions can be removed piecemeal as follows:
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
myHappn3Instance.services.security.groups.removePermission('TEST'/* group name*/, '/test/path/*' /* permission path */, 'on' /* action */)
.then(function () {
//users belonging to the TEST group can now only do "set" and "get" operations, the right to do "on" has been revoked
})
.catch(done);
});
remove a permission by upserting a group
- a group can be upserted with a set of permissions which are merged into the permissions tree, permissions that have "prohibit" actions are removed *
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, myHappn3Instance) {
var myGroup = {
name:'TEST',
permissions:{
'/test/path/*':{actions:['get', 'set']}, //allow only gets and sets to this path
'/test/allow/all':{actions:['*']} //allow all actions to this path
'/test/do/not/subscribe':{prohibit:['on']} //prohibit "on" requests to this path
},
custom_data:{//any custom data you want
test:'data'
}
};
myHappn3Instance.services.security.groups.upsertGroup(myGroup)
.then(function(upserted){
//group updated, "on" permissions to "/test/do/not/subscribe" have been deleted if they existed previously
});
});
ANONYMOUS USER
The anonymous user can be enabled using the following configuration:
var happn = require('happn-3')
var happnInstance = await happn.service.create({
secure: true,
services: {
security: {
config: {
allowAnonymousAccess: true //allow creation of anonymous user
}
}
}
});
const anonymousClient = await happn.client.create({
username: '_ANONYMOUS'
});
- the anonymous user can log in without a password, the username must just be _ANONYMOUS
- the anonymous user can be given permissions to allow for actions in the system that dont require authentication
USER PERMISSIONS
It is not necessary to use groups as the containers for permissions - permissions can also be added directly to users:
const myUser = {
username: 'test_username',
password: 'test_pwd',
permissions: {}
};
myUser.permissions['/test/path/*'] = {
actions: ['on', 'set']
};
await myHappn3Instance.services.security.users.upsertUser(testUser2, {});
//this user specifically can set and listen on /test/path/*
//you can now remove the 'on' permission:
await myHappn3Instance.services.security.users.removePermission(
addedTestuser2.username,
'/test/path/*',
'on'
)
//you can now remove the 'set' permission:
await myHappn3Instance.services.security.users.removePermission(
addedTestuser2.username,
'/test/path/*',
'set'
)
//you can now remove the 'all' permissions:
await myHappn3Instance.services.security.users.removePermission(
addedTestuser2.username,
'/test/path/*'
)
//you can re-add a permission
await myHappn3Instance.services.security.users.upsertPermission(
addedTestuser2.username,
'/test/path/*',
'set'
)
//you can list permissions for a user
const permissions = await myHappn3Instance.services.security.users.listPermissions(
addedTestuser2.username
)
//list looks like this:
[
{ action: 'set', authorized: true, path: '/test/path/*' }
]
NESTED PERMISSIONS FOR GET AND SET
by default, nested permissions are not switched on, hence a call to get or listen on 'test/path/**' will be unauthroized unless you explicitly have permissions on 'test/path/*'
Switching on nested permissions:
In the happn config, add an 'allowNestedPermissions' flag, set to true:
const Happn = require('happn-3');
let myService = await Happn.service.create({
name: 'TEST-NAME',
secure: true,
allowNestedPermissions: true,
services: {
...
}
});
Nested permissions behaviour:
If you attempt to get or listen on a path ending in '/**', your permissions on that path, and its subpaths, will be evaluated, and you will get, or listen on, all subpaths which you have access to, and only those subpaths. Note that if you do not have permissions on any subpaths, you will still get an 'unauthorized' response. In the case of events, if you have subscribed on a nested path (ending in '/**'), your subscriptions will change if your permissions change on subpaths of that path.
Nested listener behaviour example:
var myGroup = {
name:'TEST',
permissions:{
'/test/path/1':{actions:['get', 'on']},
'/test/path/2/3':{actions:['get', 'on']},
'/test/path/4/5/6':{actions:['get', 'on']}
}
};
A user who is a member of 'myGroup' and subscribes on 'test/path/**' will get events on exactly '/test/path/1', '/test/path/2/3' and '/test/path/4/5/6', and they will fire the handler for 'test/path/**' If the group subsequently loses permissions on 'test/path/1', events on that path will no longer fire the handler, and similarly if the group gains permissions on 'test/path/xyz', events on that path will fire the handler.
VOLATILE PERMISSIONS
by default, permissions are persisted via the default data provider, this is typically to a file (when using nedb) - or to a mongo database. In situations where you do not want to persist permissions between restarts, you can by adjusting the persistPermissions security config setting to false:
const Happn = require('happn-3');
let myService = await Happn.service.create({
name: 'TEST-NAME',
secure: true,
services: {
security: {
config: {
persistPermissions:false
}
},
data: {
config: {
filename: filename
}
}
}
});
*NB: *
SECURITY PROFILES
profiles can be configured to fit different session types, profiles are ordered sets of rules that match incoming sessions with specific policies, the first matching rule in the set is selected when a session is profiled, so the order they are configured in the array is important
//there are 2 default profiles that exist in secure systems - here is an example configuration
//showing how profiles can be configured for a service:
var serviceConfig = {
services:{
security: {
config: {
sessionTokenSecret:"TESTTOKENSECRET",
keyPair: {
privateKey: 'Kd9FQzddR7G6S9nJ/BK8vLF83AzOphW2lqDOQ/LjU4M=',
publicKey: 'AlHCtJlFthb359xOxR5kiBLJpfoC2ZLPLWYHN3+hdzf2'
},
profiles:[ //profiles are in an array, in descending order of priority, so if you fit more than one profile, the top profile is chosen
{
name:"web-session",
session:{
$and:[{
user:{username:{$eq:'WEB_SESSION'}},
type:{$eq:0}
}]
},
policy:{
ttl: "4 seconds",//4 seconds = 4000ms, 4 days = 1000 * 60 * 60 * 24 * 4, allow for hours/minutes
inactivity_threshold:2000//this is costly, as we need to store state on the server side
}
}, {
name:"rest-device",
session:{
$and:[{ //filter by the security properties of the session - check if this session user belongs to a specific group
user:{groups:{
"REST_DEVICES" : { $exists: true }
}},
type:{$eq:0} //token stateless
}]},
policy: {
ttl: 2000//stale after 2 seconds
}
},{
name:"trusted-device",
session:{
$and:[{ //filter by the security properties of the session, so user, groups and permissions
user:{groups:{
"TRUSTED_DEVICES" : { $exists: true }
}},
type:{$eq:1} //stateful connected device
}]},
policy: {
ttl: 2000,//stale after 2 seconds
permissions:{//permissions that the holder of this token is limited, regardless of the underlying user
'/TRUSTED_DEVICES/*':{actions: ['*']}
}
}
},{
name:"specific-device",
session:{$and:[{ //instance based mapping, so what kind of session is this?
type:{$in:[0,1]}, //any type of session
ip_address:{$eq:'127.0.0.1'}
}]},
policy: {
ttl: Infinity,//this device has this access no matter what
inactivity_threshold:Infinity,
permissions:{//this device has read-only access to a specific item
'/SPECIFIC_DEVICE/*':{actions: ['get','on']}
}
}
},
{
name:"non-reusable",
session:{$and:[{ //instance based mapping, so what kind of session is this?
user:{groups:{
"LIMITED_REUSE" : { $exists: true }
}},
type:{$in:[0,1]} //stateless or stateful
}]},
policy: {
usage_limit:2//you can only use this session call twice
}
}, {
name:"default-stateful",// this is the default underlying profile for stateful sessions
session:{
$and:[{type:{$eq:1}}]
},
policy: {
ttl: Infinity,
inactivity_threshold:Infinity
}
}, {
name:"default-stateless",// this is the default underlying profile for ws sessions
session:{
$and:[{type:{$eq:0}}]
},
policy: {
ttl: 60000 * 10,//session goes stale after 10 minutes
inactivity_threshold:Infinity
}
}, {
name:"ip address whitelist",// this ensures the _ADMIN user can only login from a whitelisted set of IP addresses (in this case locally)
session:{
$and:[{
user:{username:{$eq:'WEB_SESSION'}}
}]
},
policy: {
sourceIPWhitelist: [
'127.0.0.1',
'::ffff:127.0.0.1' //NOTE: if proxied be sure to also allow for IPV6 prefixed addresses
]
}
}
]
}
}
}
};
the test that clearly demonstrates profiles can be found here
the default policies look like this:
//stateful - so ws sessions:
{
name:"default-stateful",// this is the default underlying profile for stateful sessions
session:{
$and:[{type:{$eq:1}}]
},
policy: {
ttl: 0, //never goes stale
inactivity_threshold:Infinity
}
}
//stateless - so token based http requests (REST)
{
name:"default-stateless",// this is the default underlying profile for stateless sessions (REST)
session:{
$and:[{type:{$eq:0}}]
},
policy: {
ttl: 0, //never goes stale
inactivity_threshold:Infinity
}
}
NB NB - if no matching profile is found for an incoming session, one of the above is selected based on whether the session is stateful or stateless, there is no ttl or inactivity timeout on both policies - this means that tokens can be reused forever (unless the user in the token is deleted) rather push to default policies to your policy list which would sit above these less secure ones, with a ttl and possibly inactivity timeout
TEMPLATED PERMISSIONS
permissions can be templated to the current session using {{handlebars}} syntax, the template context is the current session and its sub-objects
code snippets have been taken from the this test
average session object looks like this:
var sessionObj = {
"id": "ec5d673b-cf28-4a0a-8a12-396f20aaaf57",//session unique id, transient
"username": "TEST [email protected]_V1DWIA5xxB",//session username
"isToken": false,//whether the session was accessed via a token or not
"permissionSetKey": "uJgq//rc4saoc1MSjqyB3KvJEUs=",//hash of user permissions for quick access lookup
"user": {
"username": "TEST [email protected]_V1DWIA5xxB",
"custom_data": {//any custom user info
"custom_field2": "custom_field2_changed",
"custom_field3": "custom_field3_changed",
"custom_field4": "custom_field4_value",
"custom_field5": "custom_field5_value",
"custom_field_forbidden": "*"
},
"groups": {//groups the user belongs to
"TEST GROUP1528278832638_V1DWIA5xxB": {
"data": {}//any data that may have been added to the group
}
}
}
}
we can save a group with templated permissions that give the user access to paths containing its username like so:
var testGroup = {
name: 'TEST GROUP' + test_id,
custom_data: {
customString: 'custom1',
customNumber: 0
}
};
testGroup.permissions = {
'/gauges/{{user.username}}': {
actions: ['*']
},
'/gauges/{{user.username}}/*': {//NB: note here how the path relates to the above session object's user.username
actions: ['*']
},
'/custom/{{user.custom_data.custom_field1}}': {//NB: note here how the path relates to the above session object
actions: ['get', 'set']
}
};
var testUser = {
username: 'TEST [email protected]' + test_id,
password: 'TEST PWD',
custom_data: {
custom_field1: 'custom_field1_value'//NB: note here how the value will match the below requests
}
};
var addedTestGroup, addedTestUser, testClient;
//assuming we have created a happn instance and assigned it to myHappnService
myHappnService.services.security.groups.upsertGroup(testGroup)
.then(function(result){
addedTestGroup = result;
return myHappnService.services.security.users.upsertUser(testUser);
})
.then(function(result){
addedTestUser = result;
return myHappnService.services.security.users.linkGroup(addedTestGroup, addedTestuser);
})
.then(function(result){
return myHappnService.services.session.localClient({
username: testUser.username,
password: 'TEST PWD'
});
})
.then(function(clientInstance){
testClient = clientInstance;
done();
})
.catch(done);
then from the testClient access the path the template resolves to like so:
var username = 'TEST [email protected]' + test_id;
testClient.on('/gauges/' + username, function(data) {
expect(data.value).to.be(1);
expect(data.gauge).to.be('logins');
done();
}, function(e) {
if (e) return done(e);
testClient.increment('/gauges/' + username, 'logins', 1, function(e) {
if (e) return done(e);
});
});
NB: permissions that resolve to context properties with * in them are ignored lest there be the chance for unauthorized promotion of the users privileges, ie:
//for the given user object with a custom_data property with a * fields
var exampleUser = {
username:'test',
password:'blah',
custom_data:{
test:'*'
}
}
//this permission will not work:
var exampleGroup = {
name:'test',
permissions:{
'users/{{user.custom_data.test}}/gauges/*':{actions:['on', 'set']}
}
}
//because the template would resolve to users/*/gauges/*, which may leak other users data
WEB PATH LEVEL SECURITY
the http/s server that happn uses can also have custom routes associated with it, when the service is run in secure mode - only people who belong to groups that are granted @HTTP permissions that match wildcard patterns for the request path can access resources on the paths, here is how we grant permissions to paths:
var happn = require('happn-3')
var happnInstance; //this will be your server instance
happn.service.create({secure:true, adminUser:{password:'testPWD'}},
function (e, instance) {
if (e)
return callback(e);
happnInstance = instance; //here it is, your server instance
var testGroup = {
name:'TEST GROUP',
custom_data:{
customString:'custom1',
customNumber:0
}
}
testGroup.permissions = {
'/@HTTP/secure/route/*':{actions:['get']},//NB - we can wildcard the path
'/@HTTP/secure/another/route/test':{actions:['put','post']}//NB - actions confirm to http verbs
};
happnInstance.services.security.upsertGroup(testGroup, {}, function(e, group){
//our group has been upserted with the right permissions
//this is how we add custom routes to the service, these routes are both available to users who belong to the 'TEST GROUP' group or the _ADMIN user (who has permissions to all routes)
happnInstance.connect.use('/secure/route/test', function(req, res, next){
res.setHeader('Content-Type', 'application/json');
res.end(JSON.stringify({"secure":"value"}));
});
happnInstance.connect.use('/secure/another/route/test', function(req, res, next){
res.setHeader('Content-Type', 'application/json');
res.end(JSON.stringify({"secure":"value"}));
});
});
});
logging in with a secure client gives us access to a token that can be used, either by embedding the token in a cookie called happn_token or a query string parameter called happn_token, if the login has happened on the browser, the happn_token is automatically set by default
//logging in with the _ADMIN user, who has permission to all web routes
var happn = require('happn-3');
happn.client.create({username:'_ADMIN', password:'testPWD'},function(e, instance) {
//the token can be derived from instance.session.token now
//here is an example of an http request using the token:
var http = require('http');
var options = {
host: '127.0.0.1',
port:55000,
path:'/secure/route/test'
}
if (use_query_string) options.path += '?happn_token=' + instance.session.token;
else options.headers = {'Cookie': ['happn_token=' + instance.session.token]}
http.request(options, function(response){
//response.statusCode should be 200;
}).end();
});
USING A BEARER TOKEN AUTHORIZATION HEADER
*A Bearer authorization token can also be used to do http requests with, as follows: *
var happn = require('happn-3');
happn.client.create({username:'_ADMIN', password:'testPWD'},function(e, instance) {
var request = require('request');
var options = {
url: 'http://127.0.0.1:55000/my/special/middleware',
};
options.headers = {'Authorization': ['Bearer ' + instance.session.token]};
request(options, function (error, response, body) {
//response happens all should be ok if the token is correct and the account is able to access the middleware resource
});
});
SECURITY OPTIONS
disableDefaultAdminNetworkConnections - this config setting prevents any logins of the default _ADMIN user via the network, thus only local (intra process) _ADMIN connections are allowed:
var happn = require('happn-3');
happn.service.create({
secure: true,
port:55002,
disableDefaultAdminNetworkConnections:true //here is our switch
}, function(e, service){
happn_client.create({
config: {
username: '_ADMIN',
password: 'happn',
port:55002
}
}, function (e, instance) {
//we will have an error here
expect(e.toString()).to.be('AccessDenied: use of _ADMIN credentials over the network is disabled');
});
});
//only clients peeled off the local process will work:
serviceInstanceLocked.services.session.localAdminClient(function(e, adminClient){
if (e) return done(e);
adminClient.get('/_SYSTEM/*', function(e, items){
expect(items.length > 0).to.be(true);//fetched system level data
});
});
//or
serviceInstanceLocked.services.session.localClient({
username:'_ADMIN',
password:'happn'
}, function(e, adminClient){
adminClient.get('/_SYSTEM/*', function(e, items){
expect(items.length > 0).to.be(true);//fetched system level data
});
});
AUTHENTICATION PROVIDER(S)
Happn-3 can now be configured with multiple, and/or custom authentication providers. By default, happn-3 will launch using the "happn" authentication provider, and work as it has always done. In order to use a different authentication provider, you must pass in either an absolute path, or an installed module name, in the security service config.
Example: Happn-3 and 2 other providers, otherAuthProvider default:
const serviceConfig = {
secure: true,
services: {
security: {
config: {
authProviders:{
otherAuthProvider: '/path/to/other/provider.js',
moduleProvider: 'moduleName'
},
defaultAuthProvider: 'otherAuthProvider'
}
}
};
var happn = require('../lib/index')
var service = happn.service;
service.create(serviceConfig, function(e, happnInst) {
//server created with three auth providers, one at /path/to/other/provider.js, one in module 'moduleName', and the standard happn provider
});
Specifying which authentication provider to use:
The authentication provider to use can be specified either by saving a user with an authType property, or by specifying the authType in the login request
Defining authType by user:
This below instance's security service is being created with the option allowUserChooseAuthProvider: false
, that disables the authType being provided on request, this means the authType must be attached to the user, requests with authType defined against a server configured with allowUserChooseAuthProvider: false
will be rejected as with an "Invalid credentials" response:
let happn = require('happn-3');
let service = await happn.service.create({
secure: true,
services: {
security: {
config: {
//NB: this means login requests will not be able to specifiy the authType
allowUserChooseAuthProvider: false,
authProviders:{
customAuthProvider: 'my-custom-auth-module',
}
}
}
}
});
A user is added with the property authType: 'customAuthProvider'
, logins for this username will automatically use the configured custom authentication provider:
await service.services.security.users.upsertUser({
username: 'test',
password: 'test',
authType: 'customAuthProvider'
});
Defining authType by request:
In order to specify which authentication provider to use, requests should add a flag, authType: 'auth-provider-name'
, to the credentials object used at login, where name is the providers name (as it appears as a key in config.authProviders).
Example to authenticate with moduleProvider:
happn.client.create({username: "user", password: "pass", authType: "moduleProvider"}, ...)
\\or
happpnInstance.services.security.login( {username: "user", password: "pass", authType: "moduleProvider"}, ...)
NB: this will only work with a server security configuration with allowUserChooseAuthProvider: undefined || true
CREATING CUSTOM AUTH PROVIDERS (AND TEMPLATE)
The file or module which contains the custom auth provider should be structured as a function which returns a class that extends the functions argument. The custom auth provider should implement at least one of the functions, __providerCredsLogin and __providerTokenLogin and will have access to the __loginOK and __loginFailed methods as well as the other methods of the base auth provider class which can be examined at ./lib/services/security/authentication/provider-base.js
Example/template auth module:
// the BaseAuthProviderClass is passed in by the system, you need to extend it in your custom auth provider
module.exports = function(BaseAuthProviderClass) {
return class AuthProvider extends BaseAuthProviderClass {
coonstructor(happn,config) {
super(happn,config)
}
static create(happn,config) {
return new AuthProvider(happn, config);
}
// called when we are logging in via standard user credentials (username, password)
__providerCredsLogin(credentials, sessionId, callback) {
// Examine credentials.username and credentials.password
// __loginOK already exists on the custom provider via inheritence of BaseAuthProviderClass
// User can be fetched from this.users.getUser(); (this.users exposed via inheritence)
if (credentials.username === 'good user') {
return this.__loginOK(credentials, user, sessionId, callback);
}
// __loginFailed already exists on the custom provider via inheritence of BaseAuthProviderClass
this.__loginFailed(credentials.username, 'ErrorMessage', new Error("failed"), callback );
}
// called when we are logging in via a token (token is present in the credentials)
__providerTokenLogin(credentials, decodedToken, sessionId, callback) {
// Examine credentials.token
//Login OK
if (credentials.token === "good token" && decodedToken.username === "good user") {
return this.__loginOK(credentials, user, sessionId, callback);
}
// Bad or stale token
this.__loginFailed(decodedToken.username, 'ErrorMessage', new Error("failed"), callback);
}
}
}
UNCONFIGURED SESSION CLEANUP
because we do not use client certificates to manage connections to a happn instance as part of the framework, sockets can be created that are not necessarily logged in, although they would be unable to do anything data-wise, these sockets could clog up allowed upgrade requests in a rate limited setup. The server, if secure, can be setup to disconnect sockets that have not logged in and have no user data attached to them within a specific period
const serviceConfig = {
secure: true,
services: {
session: {
config: {
unconfiguredSessionCleanup: {
interval: 5e3, //check every N milliseconds for unconfigured s