I'm building my first meteor app and need to be able to create a new route handler to handle an oauth callback. I've looked through server.js and found that the connect.app context is available under meteor_bootstrap. Although this doesn't seem to work:
if (Meteor.is_server) {
Meteor.startup(function () {
var app = __meteor_bootstrap__.app;
app.use('/callback',function (req,res) {
res.writeHead(404);
res.end();
return;
});
});
}
Thoughts?
The problem with this solution is that your middleware is put at the bottom of the stack. Therefore the catch-all meteor handler will always run before your "/callback"-handler.
One very hacky way to get around this (until the meteor releases their proper routing support) is to splice in your handler att the top of the stack:
__meteor_bootstrap__.app.stack.splice (0, 0, {
route: '/hello',
handle: function (req,res, next) {
res.writeHead(200, {'Content-Type': 'text/plain'});
res.end("hello world");
return;
}.future ()
});
You can achieve this with the Meteor Router smart package:
Meteor.Router.add({
'/callback': 404
})
Some of the answers are leading to routing being a no-go on the server right now without being hacky. It's a known issue, and sounds like routing is a hot item on the todo list.
Related
I want to subscribe to data on specific pages, so I put subscribe() inside router.js. I am not sure if I should enclose it inside Meteor.isClient() block. Should I? When will I ever do routing for the server-side?
Router.route('/courses/:_id', function () {
if (Meteor.isClient) {
Meteor.subscribe("comments");
}
this.render('CourseDetail', { .. });
});
Instead of putting the if (Meteor.isClient){} check inside of your router.js file, you can simply remove that check and put the file inside of the top-level client folder in your application directory. This way, you do not have to worry about your routes being processed on the server at all. In making that change, you can structure your route definition above in one of two ways:
Router.route('/courses/:id', function() {
this.wait(Meteor.subscribe('comments')); // Either this one
this.subscribe('comments').wait(); // or this one. DO NOT DO BOTH.
if(this.ready()) {
this.render();
} else {
this.render('CourseDetail');
}
});
or:
Router.route('/courses/:id', {
subscriptions: function() {
this.subscribe('comments');
},
action: function() {
this.render('CourseDetail');
}
});
Notice that the first option passes a function as the second parameter to the Router.route() function while the second option passes an object as the second parameter to the Router.route() function. Both options are perfectly valid. For information on the first option, check this out; for information on the second option, check this out.
As for when you would ever do server-side routing, this is usually done if you are setting up an HTTP request/response part of your application for external applications to access your server. Unless this is the case, you would most likely never need to worry about setting up such a thing. In the case of doing this, however, you would define your routes and put them in the top-level server folder in your application directory. for information on server-side routing, check this out.
No need to use the Meteor wrapper. Iron router has its own syntax for saying where you want your route to run.
Here is an example.
Router.route('/item', function () { var req = this.request; var res = this.response; res.end('hello from the server\n'); }, {where: 'server'});
Here is the docs site.
https://github.com/iron-meteor/iron-router/blob/devel/Guide.md
You should define the subscription in the onBeforeAction option of the iron router route.
Router.route('/courses/:_id', function () {
onBeforeAction: function () {
Meteor.subscribe("comments");
},
action: function (){
this.render('CourseDetail', { .. });
}
});
Source
I have a /user route set up, which is supposed to render the login template if the current user isn't logged in. The entire router has a waitOn that waits for the currentUser subscription to finish. The problem is that when I go to /user it simply renders the dataNotFound template instead.
Here's the snippets of code that are relevant to this situation. I've been careful to show you them in the order they're defined in my lib/router.js file.
Router.plugin('dataNotFound', {notFoundTemplate: 'notFound'});
Router.onBeforeAction(function () {
console.log(Meteor.userId())
if (!Meteor.userId()) this.render('login');
else this.next();
}, { only: ['user'] });
Router.configure({
waitOn: function () { return Meteor.subscribe('currentUser'); }
});
Router.route('/user', {
name: 'user',
template: 'userView',
data: function () { return Meteor.user(); }
});
That console.log above doesn't even ever fire. It seems to me that since it should be a reactive function that even if initially the dataNotFound is rendered, then soon after that the onBeforeAction should be fired and render the login template, right?
It's very bizarre that your console log doesn't even fire. I have a few ideas, but first want to address the last piece of your question.
The dataNotFound plugin is triggered when the data function fires on your rout. This means it is bypassing your onBeforeAction hook altogether, and not that it isn't getting there.
One thing I can think of that might be worth trying would be wrapping the 'user' route action in a if ( this.ready() ) statement:
edit:
Router.route('user', {
// Your other stuff
action: function() {
if this.ready() {
this.render();
},
The reason I suggest this is just that you are using a waitOn statement, but I'm not 100% sure how that works if you don't have a this.ready() somewhere in your route, because that's what (again, from my reading of the documentation, have not fiddled around with it) tells the router what to wait before executing. Possibly it's not waiting at all right now.
I had a problem with onBeforeAction hook after upgrading from meteor 0.9.1 to 1.
It didnt get fired when it should. Like after I log out, I enter address manually and instead of login page I can see the actual site waiting for data that never comes. onRun hook solved it, but as docs says it gets fired only once.
Router.onRun(function () {
if (!Meteor.userId()) {
this.render('login');
} else {
this.next();
}
}, { only: ['user'] });
Try swapping out Meteor.userId() with Meteor.user() in your if statement. See this link for reference on how to handle checking for the user in a before filter: http://www.manuel-schoebel.com/blog/meteorjs-iron-router-filters-before-and-after-hooks.
I found the issue here.
According to this post, thanks to Steeve Cannon. The problem is on waitOn function.
Probably when you logginOut the subscribe to currentUser will fail, and this will make your app in infinite waiting. onBeforeAction runs after waitOn
You will need to check all variables in you Publish to insure waitOn complete successfully.
Put an if statement inside the waitOn function.
waitOn: function () {
if(Meteor.userId())
return Meteor.subscribe('currentUser');
}
Refer this comment to know why this is happening: https://github.com/iron-meteor/iron-router/issues/1010#issuecomment-72210587
I have run into an issue recently where we have been told to remove the hash symbols from our Backbone applications. This presents two problems: (a) the ASP.NET routes need to handle any remotely linked URL (currently this is no problem with the hash symbols) so that we're not hitting a 404 error and (b) the proper route needs to be preserved and passed on to the client side (Backbone) application. We're currently using ASP.NET MVC5 and Web API 2 for our backend.
The setup
For an example (and test project), I've created a test project with Backbone - a simple C# ASP.NET MVC5 Web Application. It is pretty simple (here is a copy of the index.cshtml file, please ignore what is commented out as they'll be explained next):
<script type="text/javascript">
$(document).ready(function(event) {
Backbone.history.start({
//pushState: true,
//root: "/Home/Index/"
});
var Route = Backbone.Router.extend({
routes: {
"test/:id": function (event) {
$(".row").html("Hello, " + event);
},
"help": function () {
alert("help!");
}
}
});
var appRouter = new Route();
//appRouter.navigate("/test/sometext", { trigger: true });
//appRouter.navigate("/help", { trigger: true });
});
</script>
<div class="jumbotron">
<h3>Backbone PushState Test</h3>
</div>
<div class="row"></div>
Now, without pushState enabled I have no issue remote linking to this route, ie http://localhost/Home/Index#test/sometext
The result of which is that the div with a class of .row is now "Hello, sometext".
The problem
Enabling pushState will allow us to replace that pesky # in the URL with a /, ie: http://localhost/Home/Index/test/sometext. We can use the Backbone method of router.navigate("url", true); (as well as other methods) to use adjust the URL manually. However, this does not solve the problem of remote linking. So, when trying to access http://localhost/Home/Index/test/sample you just end up with the typical 404.0 error served by IIS. so, I assume that it is handled in in the RouteConfig.cs file - inside, I add a "CatchAll" route:
routes.MapRoute(
name: "CatchAll",
url: "{*clientRoute}",
defaults: new { controller = "Home", action = "Index" }
);
I also uncomment out the pushState and root attributes in the Backbone.history.start(); method:
Backbone.history.start({
pushState: true,
root: "/Home/Index/"
});
var Route = Backbone.Router.extend({
routes: {
"test/:id": function (event) {
$(".row").html("Hello, " + event);
},
"help": function () {
alert("help!");
}
}
});
var appRouter = new Route();
//appRouter.navigate("/test/sometext", { trigger: true });
//appRouter.navigate("/help", { trigger: true });
This allows me to at least let get past the 404.0 page when linking to these routes - which is good. However, none of the routes actually "trigger" when I head to them. After attempting to debug them in Chrome, Firefox, and IE11 I notice that none of the events fire. However, if I manually navigate to them using appRouter.navigate("/help", { trigger: true }); the routes are caught and events fired.
I'm at a loss at this point as to where I should start troubleshooting next. I've placed my Javascript inside of the $(document).ready() event as well as the window.onload event also (as well as not inside of an event); none of these correct the issue. Can anyone offer advice on where to look next?
You simply have to move Backbone.history.start after the "new Route" line.
var Route = Backbone.Router.extend({
routes: {
"test/:id": function (event) {
$(".row").html("Hello, " + event);
},
"help": function () {
alert("help!");
}
}
});
var appRouter = new Route();
Backbone.history.start({
pushState: true,
root: "/Home/Index/"
});
Make sure you go to ".../Home/Index/help". If it doesn't work, try temporarily removing the root and go to ".../help" to see if the root is the problem.
If you still have troubles, set a js breakpoint in Backbone.History.loadUrl on the "return" line. It is called from the final line of History.start to execute the current browser url on page load. "this.matchRoot()" must pass then, "fragment" is matched against each "route" or regexp string in "this.handlers". You can see why or why not the browser url matches the route regexps.
To set to the js breakpoint, press F12 in the browser to open the dev console, press Ctrl-O or Ctrl-P to open a js file, then type the name of the backbone js file. Then search for "loadUrl:". You can also search for "Router =" to find the start of the router class definition (same as for "View =" and "Model =" to find the backbone view/model implementation code). I find it quite useful to look at the backbone code when I have a question like this. It is surprisingly readable and what better place to get answers?
If your js files happen to be minified/compressed, preferably turn this off. Alternately you can try the browser unminify option. In Chrome this is the "{}" button or "pretty print". Then the js code is not all on 1 line and you can set breakpoints. But the function and variable names may still be mangled.
I have solved my own problem using what feels to be "hackish", via the following. If anyone can submit a better response it would be appreciated!
My Solution:
I globally override the default Backbone.Router.intilaize method (it is empty) with the following:
$(document).ready(function (event) {
var _root = "/Home/Index/";
_.extend(Backbone.Router.prototype, {
initialize: function () {
/* check for route & navigate to it */
var pathName = window.location.pathname;
var route = pathName.split(_root)[1];
if (route != undefined && route != "") {
route = "/" + route;
this.navigate("", { trigger: false });
this.navigate(route, { trigger: true });
}
}
});
});
I'm using laika for testing and the meteor-router package for routing. I want to do tests that navigate to some page, fill a form, submit it and check for a success message, but I'm stuck on the navigation part. This was my first attempt:
var assert = require('assert');
suite('Router', function() {
test('navigate', function(done, server, client) {
client.eval(function() {
Meteor.Router.to('test');
var title = $('h1').text();
emit('title', title);
})
.once('title', function(title) {
assert.equal(title, 'Test');
done();
});
});
});
This doesn't work because Meteor.Router.to doesn't have a callback and I don't know how to execute the next line when the new page is loaded.
I tried also with something like this
var page = require('webpage').create();
page.open('http://localhost:3000/test', function () {
...
}
but I got the error Error: Cannot find module 'webpage'
Edit
I'm moving to iron router, so any answer with that also will be helpful.
I had the same problem. I needed to navigate to some page before running my tests. I'm using iron router as well. I figured you can't just execute Router.go('foo') and that's it. You need to wait until the actual routing took place. Fortunately the router exposes a method Router.current() which is a reactive data source that will change as soon as your page is ready. So, in order to navigate to a specific route before running my tests, I firstly run the following code block:
// route to /some/path
client.evalSync(function() {
// react on route change
Deps.autorun(function() {
if (Router.current().path == '/some/path') {
emit('return');
this.stop();
}
});
Router.go('/some/path');
});
Since this is within an evalSync()everything that follows this block will be executed after the routing has finished.
Hope this helps.
Laika now includes a waitForDOM() function you can set up to wait for a specific DOM element to appear, which in this case would be an element in the page you're loading.
client.eval(function() {
Router.go( 'test' );
waitForDOM( 'h1', function() {
var title = $('h1').text();
emit( 'title', title );
});
});
The first parameter is a jQuery selector.
I have the following:
function activate(routeData) {
// make dataservice call, evaluate results here. If condition is met, reroute:
if (true){
router.navigateTo("#/someRoute");
}
alert ("should not be shown");
}
The alert is getting hit however, and then the view changes.
How do I fully navigate away from the current item and prevent any further code in that vm from being hit?
Update:
I tried using guardroute but I have to activate the viewModel to call the dataservice that returns the data that determines whether or not I should re-route. Using guardroute totally prevents the dataservice from getting called (since nothing in the activate function will get hit).
I also tried returning from the if block but this still loads the view / viewAttached / etc so the UX is glitchy.
The following worked for me in Durandal 2.0:
canActivate: function() {
if(condition)
return {redirect: 'otherRoute'};
return true;
}
activate: // Do your stuff
It's mentioned in the documentation: http://durandaljs.com/documentation/Using-The-Router.html
Here's #EisenbergEffect answer to a quite similar discussion in google groups.
Implement canActivate on your view model. Return a promise of false,
then chain with a redirect.
You might want to give #JosepfGabriel's example (discussion) a try in Durandal 1.2. Check the correct router syntax for your Durandal version, you might have to substitute it with something like router.navigateTo("#/YourHash", 'replace').
canActivate: function () {
return system.defer(function (dfd) {
//if step 2 has a problem
dfd.resolve(false);
})
.promise()
.then(function () { router.navigate("wizard/step1", { trigger: true, replace: true }); });
}
However this is NOT working in Durandal 2.0 and there's a feature request https://github.com/BlueSpire/Durandal/issues/203 for it.
You can't call redirect into the active method.
You can override the guardRoute method from router, to implement redirections.
You can do somehting like that:
router.guardRoute= function(routeInfo, params, instance){
if(someConditios){
return '#/someRoute'
}
}
You can return a promise, true, false, the route to redirect... You can find more information about that in the next link: http://durandaljs.com/documentation/Router/
Rainer's answer was pretty good and works for me adding this small fix.
Inside the then() block simply call the navigation like this
setTimeout(function() { router.navigateTo('#/YOUR DESTINATION'); }, 200);
that should fix your problem. The setTimeout does the trick. Without it the newly navigated page catches the old NavigationCancel from the previous one.
Adding a return in your if (true) block should fix this.
function activate(routeData) {
if (true){
router.navigateTo("#/someRoute");
return;
}
alert ("should not be shown");
}