Develop Reference

Implementing RabbitMQ in Chrome Kiosk/Extension/Apps and Web Browsers

I'm developing internal applications that require push notifications but we cannot use outside services. I have started to work with RabbitMQ and have it working flawlessly inside of .NET Core. When trying to implement it the same thing with javascript I am not getting the same results.
I developed test clients in C#. I developed a client in javascript. I can make a successful connection but data is not arriving.
In C# I am using:
string e = Console.ReadLine();
Console.WriteLine("Enter a message (blank for test msg)");
string message = Console.ReadLine();
var factory = new ConnectionFactory() { HostName = "10.222.2.160" };
factory.UserName = "Test";
factory.Password = "TestPassword";
factory.VirtualHost = "/";
using (var connection = factory.CreateConnection("TestChannel"))
using (var channel = connection.CreateModel())
{
var body = Encoding.UTF8.GetBytes(message);
channel.BasicPublish(exchange: e,
routingKey: "",
basicProperties: null,
body: body);
Console.WriteLine(" [x] Sent {0}", message);
}
In Javascript:
var wsbroker = "10.222.2.160"; // mqtt websocket enabled broker
var wsport = 15675; // port for above
var client = new Paho.MQTT.Client(wsbroker, wsport, "/ws/",
"test");
client.onConnectionLost = function (responseObject) {
console.log("CONNECTION LOST - " + responseObject.errorMessage);
};
client.onMessageArrived = function (message) {
console.log("RECEIVE ON " + message.destinationName + " PAYLOAD " + message.payloadString);
};
var options = {
userName: "Test",
password: "TestPassword",
timeout: 3,
keepAliveInterval: 30,
onSuccess: function () {
console.log("CONNECTION SUCCESS");
client.subscribe('test', { qos: 1 });
},
onFailure: function (message) {
console.log("CONNECTION FAILURE - " + message.errorMessage);
}
};
if (location.protocol == "https:") {
options.useSSL = true;
}
console.log("CONNECT TO " + wsbroker + ":" + wsport);
client.connect(options);
I need to be able to connect to rabbitmq from javascript (non-node, chrome kiosk application/chrome extension). However, I'm not sure I am "understanding" RabbitMQ. Pointing me in the right direction would help a girl out. Thanks!

You're publishing without a routing key here:
channel.BasicPublish(exchange: e,
routingKey: "",
basicProperties: null,
body: body);
Ensure that the test queue exists, then change routingKey to test in your publisher, and use the exchange named amq.direct.
You should read the RabbitMQ introduction available here to get familiar with how exchanges, queues, routing keys and bindings interact:
https://www.cloudamqp.com/blog/2015-05-18-part1-rabbitmq-for-beginners-what-is-rabbitmq.html
NOTE: the RabbitMQ team monitors the rabbitmq-users mailing list and only sometimes answers questions on StackOverflow.

My twitter bots are posting for each other, and I checked the code. It doesn't make sense

I made a twitter bot for my personal Twitter account last night, which turned out perfectly. I even deployed it to Heroku and it worked fine. However, today I built one for my business too, and when I deployed it on Heroku, the two bots started tweeting the same thing.
Here's my code for my personal bot:
console.log('bot is starting');
var Twit = require('twit');
var config = require('./businessconfig');
var T = new Twit(config);
//Setting up stream
var stream = T.stream('user');
//Anytime someone follows me
stream.on('follow', followed);
function followed(event) {
console.log('Follow event initiated');
var name = event.source.name;
var screenName = event.source.screen_name;
tweetIt('#' + screenName + ' , thanks for following! #twitterbot');
}
function tweetIt(txt) {
var tweet = {
status: txt
}
T.post('statuses/update', tweet, tweeted);
function tweeted(err, data, response){
if (err) {
console.log(err);
}
else {
console.log("Your tweet was posted!");
}
}
}
This is in a file titled "business bot.js." The "businessconfig file is my Twitter API key stuff. Here's my personal bot code:
console.log('bot is starting');
var Twit = require('twit');
var config = require('./myconfig');
var T = new Twit(config);
//Setting up stream
var stream = T.stream('user');
//Anytime someone follows me
stream.on('follow', followed);
function followed(event) {
console.log('Follow event initiated');
var name = event.source.name;
var screenName = event.source.screen_name;
tweetIt('#' + screenName + ' , thanks for following! #twitterbot');
}
function tweetIt(txt) {
var tweet = {
status: txt
}
T.post('statuses/update', tweet, tweeted);
function tweeted(err, data, response){
if (err) {
console.log(err);
}
else {
console.log("Your tweet was posted!");
}
}
}
It is the same code as the business bot, but they are in different folders and have different package.json files. When I run the node processes, they begin to post with each other. Why do they have a relationship? For instance, if someone follows my personal account, the personal account and business account both tweet "#... Thanks for following! #twitterbot." It doesn't make sense since I made them separately and even changed the file names multiple times. I checked the API keys and they are different. Why do they talk together? Thanks.

ADAL JS not attaching user token while invoking WebApi

I am using ADAL JS for authenticating the users against Azure AD. And as I am new to ADAL JS, I started reading with following articles, which I find very informative:
Introducing ADAL JS v1
ADAL JavaScript and AngularJS – Deep Dive
After reading the articles, I had the impression that ADAL JS intercepts the service calls and if the service url is registered as one of the endpoint in AuthenticationContext configuration, it attaches the JWT token as Authentication Bearer information.
However, I found the same is not happening in my case. And after some digging, it seemed to me that it is only possible, if adal-angular counter part is also used, which I am not using currently, simply because my web application is not based on Angular.
Please let me know if my understanding is correct or not. If I need to add the bearer information explicitly, the same can be done, but I am more concerned whether I am missing some out-of-the-box facility or not.
Additional Details: My present configuration looks like following:
private endpoints: any = {
"https://myhost/api": "here_goes_client_id"
}
...
private config: any;
private authContext: any = undefined;
....
this.config = {
tenant: "my_tenant.onmicrosoft.com",
clientId: "client_id_of_app_in_tenant_ad",
postLogoutRedirectUri: window.location.origin,
cacheLocation: "sessionStorage",
endpoints: this.endpoints
};
this.authContext = new (window["AuthenticationContext"])(this.config);
Also on server-side (WebApi), Authentication configuration (Startup.Auth) is as follows:
public void ConfigureOAuth(IAppBuilder app, HttpConfiguration httpConfig)
{
app.UseWindowsAzureActiveDirectoryBearerAuthentication(
new WindowsAzureActiveDirectoryBearerAuthenticationOptions
{
Tenant = "my_tenant.onmicrosoft.com",
TokenValidationParameters = new TokenValidationParameters
{
ValidAudience = "client_id_of_app_in_tenant_ad"
}
});
}
However, the Authorization is always null in request.Headers.
UPDATE: It seems that the same applies for auto-renewal of tokens as well; when used in conjunction with adal-angular, the renewal of token works seamlessly by calling AuthenticationContext.acquireToken(resource, callback) under the hood. Please correct me if I am wrong.

After reading the articles, I had the impression that ADAL JS intercepts the service calls and if the service url is registered as one of the endpoint in AuthenticationContext configuration, it attaches the JWT token as Authentication Bearer information.
This will work only if your application is angular based. As you mentioned, the logic for this lives in adal-angular.
If, however, you want to stick to pure JS, you will not get the automatic "get-access-token-and-attach-it-to-header" support. You can use acquireToken(resource, callback api to get a token for the endpoint. But you will have to do some work in the controller that is sending the request to the api.
This might give you some idea: https://github.com/Azure-Samples/active-directory-javascript-singlepageapp-dotnet-webapi/blob/master/TodoSPA/App/Scripts/Ctrls/todoListCtrl.js. This sample does not uses angular.

ADAL.JS is incompatible with v2.0 implicit flow. I could not get it working since I set my project up recently and don't think projects are backwards compatible.
This was very confusing and took me a long time to figure out that I was mixing up the versions, and can't use ADAL.JS with v2.0. Once I removed it, things went much smoother, just did a couple of XHR requests and a popup window, no magic actually required!
Here is code for v2:
function testNoADAL() {
var clientId = "..guid..";
var redirectUrl = "..your one.."
var authServer = "https://login.microsoftonline.com/common/oauth2/v2.0/authorize?";
var responseType = "token";
var stateParam = Math.random() * new Date().getTime();
var authUrl = authServer +
"response_type=" + encodeURI(responseType) +
"&client_id=" + encodeURI(clientId) +
"&scope=" + encodeURI("https://outlook.office.com/Mail.ReadWrite") +
"&redirect_uri=" + encodeURI(redirectUrl) +
"&state=" + stateParam;
var popupWindow = window.open(authUrl, "Login", 'width=' + 300 + ', height=' + 600 + ', top=' + 10 + ', left=' + 10 + ',location=no,toolbar=yes');
if (popupWindow.focus) {
popupWindow.focus();
}
}
Note: redirectUrl will appear in popup window, needs to have code in it to pass location hash, such as this:
<script>window.opener.processMicrosoftAuthResultUrl(location.hash);window.close();</script>
function processMicrosoftAuthResultUrl(hash) {
if (hash.indexOf("#") == 0) {
hash = hash.substr(1);
}
var obj = getUrlParameters(hash);
if (obj.error) {
if (obj.error == "invalid_resource") {
errorDialog("Your Office 365 needs to be configured to enable access to Outlook Mail.");
} else {
errorDialog("ADAL: " + obj.error_description);
}
} else {
if (obj.access_token) {
console.log("ADAL got access token!");
var token = obj.access_token;
var url = "https://outlook.office.com/api/v2.0/me/MailFolders/Inbox/messages";
$.ajax({
type: "GET",
url: url,
headers: {
'Authorization': 'Bearer ' + token,
},
}).done(function (data) {
console.log("got data!", data);
var message = "Your latest email is: " + data.value[0].Subject + " from " + data.value[0].From.EmailAddress.Name+ " on " + df_FmtDateTime(new Date(data.value[0].ReceivedDateTime));
alertDialog(message);
}).fail(function () {
console.error('Error getting todo list data')
});
}
}
}
function getUrlParameters(url) {
// get querystring and turn it into an object
if (!url) return {};
if (url.indexOf("?") > -1) {
url = url.split("?")[1];
}
if (url.indexOf("#") > -1) {
url = url.split("#")[0];
}
if (!url) return {};
url = url.split('&')
var b = {};
for (var i = 0; i < url.length; ++i) {
var p = url[i].split('=', 2);
if (p.length == 1) {
b[p[0]] = "";
} else {
b[decodeURIComponent(p[0])] = decodeURIComponent(p[1].replace(/\+/g, " "));
}
}
return b;
}

How do I close a sockjs connection on the server side?

So, every time I refresh the page, it seems like sockjs is creating a new connection.
I am saving every message to my mongodb on every channel.onmessage, so if I refresh my page 7 times and send a message, I would save 7 messages of the same content into my mongodb.
This is very problematic because when I retrieve those messages when I go into the chat room, to see the log, I would see bunch of duplicate messages.
I want to keep track of all connections that are 'active', and if a user tries to make another connections, I want to be able to force close on the old one from the server side, so there is only 1 connection listening to each message at a time
How do I do this ?
var connections = {};
//creating the sockjs server
var chat = sockjs.createServer();
//installing handlers for sockjs server instance, with the same url as client
chat.installHandlers(server, {prefix:'/chat/private'});
var multiplexer = new multiplexServer.MultiplexServer(chat);
var configChannel = function (channelId, userId, userName){
var channel = multiplexer.registerChannel(channelId);
channel.on('connection', function (conn) {
// console.log('connection');
console.log(connections);
connections[channelId] = connections[channelId] || {};
if (connections[channelId][userId]) {
//want to close the extra connection
} else {
connections[channelId][userId] = conn;
}
// }
// if (channels[channelId][userId]) {
// conn = channels[channelId][userId];
// } else {
// channels[channelId][userId] = conn;
// }
// console.log('accessing channel! ', channels[channelId]);
conn.on('new user', function (data, message) {
console.log('new user! ', data, message);
});
// var number = connections.length;
conn.on('data', function(message) {
var messageObj = JSON.parse(message);
handler.saveMessage(messageObj.channelId, messageObj.user, messageObj.message);
console.log('received the message, ', messageObj.message);
conn.write(JSON.stringify({channelId: messageObj.channelId, user: messageObj.user, message: messageObj.message }));
});
conn.on('close', function() {
conn.write(userName + ' has disconnected');
});
});
return channel;
};

Simply use .close:
if (connections[channelId][userId]) {
// want to close the extra connection
connections[channelId][userId].close();
} else {
connections[channelId][userId] = conn;
}

Creating a map of ids to sockets and vice versa in Node.js

I'm trying to manage a bunch of socket connections. My app is basically an http server that receives posts and passes these along to a socket. When clients open a socket connection, they send a connect message with an id:
{"m":"connect","id":"1"}
The app then saves this id and socket in the id2socket and socket2id maps. On disconnect, the socket/id pair is deleted from the maps.
A post will also contain an id, which indicates the post data should be sent to the socket with that id.
That's great, and this works fine for a single open socket. However, when I have more than one socket open, and then I close a socket, that disconnect wipes everything from the map. I think my understanding of sockets in node is incomplete- is there only a single socket object that is used in the callback? Is there a better way to manage my open socket connections and ids?
start server:
>>node server.js
TCP server listening on 127.0.0.1:5280
HTTP server listening on 127.0.0.1:9002
telnet in:
>>telnet localhost 5280
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
{"m":"connect","id":"123"}
{"m":"connect","id":"123","success":"true"}
server after connection:
>>Connection from 127.0.0.1:57572
received data: {"m":"connect","id":"123"}
id: 1
m: connect
associating uid 1 with socket [object Object]
do a post:
python post.py {"foo":"bar"}
So this works fine for several open sockets (as long as 1 device is id 123, server has this hardwired for now). However, as soon as you close one connection all the socket connections are removed from the map.
Here's my code:
python script to do post:
import sys
import json
import httplib, urllib, urllib2
values = json.loads('{"foo":"bar"}')
headers = {"Content-type": "application/json"}
conn = httplib.HTTPConnection('127.0.0.1', 9002)
headers = {"Content-type": "application/json"}
conn.request("POST", "", json.dumps(values), headers)
response = conn.getresponse()
print "response.status: "+response.status
print "response.reason: "+response.reason
print "response.read: "+response.read()
conn.close()
node server (http and tcp), hardwired to send data to device '123' on post:
var net = require('net'); // tcp-server
var http = require("http"); // http-server
var qs = require('querystring'); // http-post
// Map of sockets to devices
var id2socket = new Object;
var socket2id = new Object;
// Setup a tcp server
var server_plug = net.createServer(function(socket) {
// Event handlers
socket.addListener("connect", function(conn) {
console.log("Connection from " + socket.remoteAddress + ":" + socket.remotePort );
});
socket.addListener("data", function(data) {
console.log("received data: " + data);
try {
request = JSON.parse(data);
response = request;
if(request.m !== undefined && request['id'] !== undefined){ // hack on 'id', id is js obj property
console.log("id: "+request['id']);
console.log("m: "+request.m);
if(request.m == 'connect'){
console.log("associating uid " + request['id'] + " with socket " + socket);
id2socket[request['id']] = socket;
socket2id[socket] = request['id'];
response.success = 'true';
} else {
response.success = 'true';
}
}
socket.write(JSON.stringify(response));
} catch (SyntaxError) {
console.log('Invalid JSON:' + data);
socket.write('{"success":"false","response":"invalid JSON"}');
}
});
socket.on('end', function() {
id = socket2id[socket]
console.log("socket disconnect by id " + id);
// wipe out the stored info
console.log("removing from map socket:"+socket+" id:"+id);
delete id2socket[id];
delete socket2id[socket];
});
socket.on('timeout', function() {
console.log('socket timeout');
});
});
// Setup http server
var server_http = http.createServer(
// Function to handle http:post requests, need two parts to it
// http://jnjnjn.com/113/node-js-for-noobs-grabbing-post-content/
function onRequest(request, response) {
request.setEncoding("utf8");
request.addListener("data", function(chunk) {
request.content += chunk;
});
request.addListener("end", function() {
console.log("post received!");
//console.log("Request received: "+request.content);
if (request.method == 'POST') {
//var json = qs.parse(request.content);
//console.log("Post: "+json);
// HACK TO TEST STUFF:
// send a message to one of the open sockets
try {
var socket = id2socket['123']; //hardwired
socket.write('{"m":"post"}');
} catch (Error) {
console.log("Cannot find socket with id "+'123');
}
}
});
}
);
// Fire up the servers
var HOST = '127.0.0.1';
var PORT = 5280;
var PORT2 = 9002;
server_plug.listen(PORT, HOST);
console.log("TCP server listening on "+HOST+":"+PORT);
server_http.listen(PORT2);
console.log("HTTP server listening on "+HOST+":"+PORT2);

Objects only take strings as keys for their properties. As your log shows, a socket object is converted into the string "[object Object]". As a result, socket #2 overwrites the id from socket #1 in the object, because all sockets are converted into the same string key. So, there is only one property in the object at all times, because all sockets come down to the same key. When you try to remove the id for socket #2, the single property is deleted and the object is empty.
You seem to want a custom property for each separate socket when used as a key. You can use WeakMaps for this. WeakMaps do allow objects as keys (as opposed to string-only keys), but as they're relatively new they may contain bugs at the moment.
(Note that the id2socket map can just be a plain object, because numbers are converted into strings just fine, and each number has its own, distinct string representation*.)
Using WeakMaps is as follows:
var socket2id = new WeakMap; // as if you were doing: var socket2id = {};
socket2id.set(socket, id); // as if you were doing: socket2id[socket] = id;
socket2id.get(socket); // as if you were doing: socket2id[socket];
socket2id.delete(socket); // as if you were doing: delete socket2id[socket];
Make sure to run with node --harmony (>= 0.7) or node --harmony_weakmaps (<= 0.6).
* 0 and -0 are exceptions, but you shouldn't be using -0 anyway because 0 === -0, so it's difficult to differ between them.