Carefully Designed Client Side Router for Meteor
JavaScript
Latest commit 618b8eb Jan 11, 2017 @arunoda arunoda committed on GitHub Merge pull request #711 from yehchuonglong/readme-typo
fix typo on README.md

README.md

FlowRouter Build Status Stories in Ready

Carefully Designed Client Side Router for Meteor.

FlowRouter is a very simple router for Meteor. It does routing for client-side apps and does not handle rendering itself.

It exposes a great API for changing the URL and reactively getting data from the URL. However, inside the router, it's not reactive. Most importantly, FlowRouter is designed with performance in mind and it focuses on what it does best: routing.

We've released 2.0 and follow this migration guide if you are already using FlowRouter.

TOC

Meteor Routing Guide

Meteor Routing Guide is a completed guide into routing and related topics in Meteor. It talks about how to use FlowRouter properly and use it with Blaze and React. It also shows how to manage subscriptions and implement auth logic in the view layer.

Meteor Routing Guide

Getting Started

Add FlowRouter to your app:

meteor add kadira:flow-router

Let's write our first route (add this file to lib/router.js):

FlowRouter.route('/blog/:postId', {
    action: function(params, queryParams) {
        console.log("Yeah! We are on the post:", params.postId);
    }
});

Then visit /blog/my-post-id from the browser or invoke the following command from the browser console:

FlowRouter.go('/blog/my-post-id');

Then you can see some messages printed in the console.

Routes Definition

FlowRouter routes are very simple and based on the syntax of path-to-regexp which is used in both Express and iron:router.

Here's the syntax for a simple route:

FlowRouter.route('/blog/:postId', {
    // do some action for this route
    action: function(params, queryParams) {
        console.log("Params:", params);
        console.log("Query Params:", queryParams);
    },

    name: "<name for the route>" // optional
});

So, this route will be activated when you visit a url like below:

FlowRouter.go('/blog/my-post?comments=on&color=dark');

After you've visit the route, this will be printed in the console:

Params: {postId: "my-post"}
Query Params: {comments: "on", color: "dark"}

For a single interaction, the router only runs once. That means, after you've visit a route, first it will call triggers, then subscriptions and finally action. After that happens, none of those methods will be called again for that route visit.

You can define routes anywhere in the client directory. But, we recommend to add them in the lib directory. Then fast-render can detect subscriptions and send them for you (we'll talk about this is a moment).

Group Routes

You can group routes for better route organization. Here's an example:

var adminRoutes = FlowRouter.group({
  prefix: '/admin',
  name: 'admin',
  triggersEnter: [function(context, redirect) {
    console.log('running group triggers');
  }]
});

// handling /admin route
adminRoutes.route('/', {
  action: function() {
    BlazeLayout.render('componentLayout', {content: 'admin'});
  },
  triggersEnter: [function(context, redirect) {
    console.log('running /admin trigger');
  }]
});

// handling /admin/posts
adminRoutes.route('/posts', {
  action: function() {
    BlazeLayout.render('componentLayout', {content: 'posts'});
  }
});

All of the options for the FlowRouter.group() are optional.

You can even have nested group routes as shown below:

var adminRoutes = FlowRouter.group({
    prefix: "/admin",
    name: "admin"
});

var superAdminRoutes = adminRoutes.group({
    prefix: "/super",
    name: "superadmin"
});

// handling /admin/super/post
superAdminRoutes.route('/post', {
    action: function() {

    }
});

You can determine which group the current route is in using:

FlowRouter.current().route.group.name

This can be useful for determining if the current route is in a specific group (e.g. admin, public, loggedIn) without needing to use prefixes if you don't want to. If it's a nested group, you can get the parent group's name with:

FlowRouter.current().route.group.parent.name

As with all current route properties, these are not reactive, but can be combined with FlowRouter.watchPathChange() to get group names reactively.

Rendering and Layout Management

FlowRouter does not handle rendering or layout management. For that, you can use:

Then you can invoke the layout manager inside the action method in the router.

FlowRouter.route('/blog/:postId', {
    action: function(params) {
        BlazeLayout.render("mainLayout", {area: "blog"});
    }
});

Triggers

Triggers are the way FlowRouter allows you to perform tasks before you enter into a route and after you exit from a route.

Defining triggers for a route

Here's how you can define triggers for a route:

FlowRouter.route('/home', {
  // calls just before the action
  triggersEnter: [trackRouteEntry],
  action: function() {
    // do something you like
  },
  // calls when we decide to move to another route
  // but calls before the next route started
  triggersExit: [trackRouteClose]
});

function trackRouteEntry(context) {
  // context is the output of `FlowRouter.current()`
  Mixpanel.track("visit-to-home", context.queryParams);
}

function trackRouteClose(context) {
  Mixpanel.track("move-from-home", context.queryParams);
}

Defining triggers for a group route

This is how you can define triggers on a group definition.

var adminRoutes = FlowRouter.group({
  prefix: '/admin',
  triggersEnter: [trackRouteEntry],
  triggersExit: [trackRouteEntry]
});

You can add triggers to individual routes in the group too.

Defining Triggers Globally

You can also define triggers globally. Here's how to do it:

FlowRouter.triggers.enter([cb1, cb2]);
FlowRouter.triggers.exit([cb1, cb2]);

// filtering
FlowRouter.triggers.enter([trackRouteEntry], {only: ["home"]});
FlowRouter.triggers.exit([trackRouteExit], {except: ["home"]});

As you can see from the last two examples, you can filter routes using the only or except keywords. But, you can't use both only and except at once.

If you'd like to learn more about triggers and design decisions, visit here.

Redirecting With Triggers

You can redirect to a different route using triggers. You can do it from both enter and exit triggers. See how to do it:

FlowRouter.route('/', {
  triggersEnter: [function(context, redirect) {
    redirect('/some-other-path');
  }],
  action: function(_params) {
    throw new Error("this should not get called");
  }
});

Every trigger callback comes with a second argument: a function you can use to redirect to a different route. Redirect also has few properties to make sure it's not blocking the router.

  • redirect must be called with an URL
  • redirect must be called within the same event loop cycle (no async or called inside a Tracker)
  • redirect cannot be called multiple times

Check this PR to learn more about our redirect API.

Stopping the Callback With Triggers

In some cases, you may need to stop the route callback from firing using triggers. You can do this in before triggers, using the third argument: the stop function. For example, you can check the prefix and if it fails, show the notFound layout and stop before the action fires.

var localeGroup = FlowRouter.group({
  prefix: '/:locale?',
  triggersEnter: [localeCheck]
});

localeGroup.route('/login', {
  action: function (params, queryParams) {
    BlazeLayout.render('componentLayout', {content: 'login'});
  }
});

function localeCheck(context, redirect, stop) {
  var locale = context.params.locale;

  if (locale !== undefined && locale !== 'fr') {
    BlazeLayout.render('notFound');
    stop();
  }
}

Note: When using the stop function, you should always pass the second redirect argument, even if you won't use it.

Not Found Routes

You can configure Not Found routes like this:

FlowRouter.notFound = {
    // Subscriptions registered here don't have Fast Render support.
    subscriptions: function() {

    },
    action: function() {

    }
};

API

FlowRouter has a rich API to help you to navigate the router and reactively get information from the router.

FlowRouter.getParam(paramName);

Reactive function which you can use to get a parameter from the URL.

// route def: /apps/:appId
// url: /apps/this-is-my-app

var appId = FlowRouter.getParam("appId");
console.log(appId); // prints "this-is-my-app"

FlowRouter.getQueryParam(queryStringKey);

Reactive function which you can use to get a value from the queryString.

// route def: /apps/:appId
// url: /apps/this-is-my-app?show=yes&color=red

var color = FlowRouter.getQueryParam("color");
console.log(color); // prints "red"

FlowRouter.path(pathDef, params, queryParams)

Generate a path from a path definition. Both params and queryParams are optional.

Special characters in params and queryParams will be URL encoded.

var pathDef = "/blog/:cat/:id";
var params = {cat: "met eor", id: "abc"};
var queryParams = {show: "y+e=s", color: "black"};

var path = FlowRouter.path(pathDef, params, queryParams);
console.log(path); // prints "/blog/met%20eor/abc?show=y%2Be%3Ds&color=black"

If there are no params or queryParams, this will simply return the pathDef as it is.

Using Route name instead of the pathDef

You can also use the route's name instead of the pathDef. Then, FlowRouter will pick the pathDef from the given route. See the following example:

FlowRouter.route("/blog/:cat/:id", {
    name: "blogPostRoute",
    action: function(params) {
        //...
    }
})

var params = {cat: "meteor", id: "abc"};
var queryParams = {show: "yes", color: "black"};

var path = FlowRouter.path("blogPostRoute", params, queryParams);
console.log(path); // prints "/blog/meteor/abc?show=yes&color=black"

FlowRouter.go(pathDef, params, queryParams);

This will get the path via FlowRouter.path based on the arguments and re-route to that path.

You can call FlowRouter.go like this as well:

FlowRouter.go("/blog");

FlowRouter.url(pathDef, params, queryParams)

Just like FlowRouter.path, but gives the absolute url. (Uses Meteor.absoluteUrl behind the scenes.)

FlowRouter.setParams(newParams)

This will change the current params with the newParams and re-route to the new path.

// route def: /apps/:appId
// url: /apps/this-is-my-app?show=yes&color=red

FlowRouter.setParams({appId: "new-id"});
// Then the user will be redirected to the following path
//      /apps/new-id?show=yes&color=red

FlowRouter.setQueryParams(newQueryParams)

Just like FlowRouter.setParams, but for queryString params.

To remove a query param set it to null like below:

FlowRouter.setQueryParams({paramToRemove: null});

FlowRouter.getRouteName()

To get the name of the route reactively.

Tracker.autorun(function() {
  var routeName = FlowRouter.getRouteName();
  console.log("Current route name is: ", routeName);
});

FlowRouter.current()

Get the current state of the router. This API is not reactive. If you need to watch the changes in the path simply use FlowRouter.watchPathChange().

This gives an object like this:

// route def: /apps/:appId
// url: /apps/this-is-my-app?show=yes&color=red

var current = FlowRouter.current();
console.log(current);

// prints following object
// {
//     path: "/apps/this-is-my-app?show=yes&color=red",
//     params: {appId: "this-is-my-app"},
//     queryParams: {show: "yes", color: "red"}
//     route: {pathDef: "/apps/:appId", name: "name-of-the-route"}
// }

FlowRouter.watchPathChange()

Reactively watch the changes in the path. If you need to simply get the params or queryParams use dedicated APIs like FlowRouter.getQueryParam().

Tracker.autorun(function() {
  FlowRouter.watchPathChange();
  var currentContext = FlowRouter.current();
  // do anything with the current context
  // or anything you wish
});

FlowRouter.withReplaceState(fn)

Normally, all the route changes made via APIs like FlowRouter.go and FlowRouter.setParams() add a URL item to the browser history. For example, run the following code:

FlowRouter.setParams({id: "the-id-1"});
FlowRouter.setParams({id: "the-id-2"});
FlowRouter.setParams({id: "the-id-3"});

Now you can hit the back button of your browser two times. This is normal behavior since users may click the back button and expect to see the previous state of the app.

But sometimes, this is not something you want. You don't need to pollute the browser history. Then, you can use the following syntax.

FlowRouter.withReplaceState(function() {
  FlowRouter.setParams({id: "the-id-1"});
  FlowRouter.setParams({id: "the-id-2"});
  FlowRouter.setParams({id: "the-id-3"});
});

Now, there is no item in the browser history. Just like FlowRouter.setParams, you can use any FlowRouter API inside FlowRouter.withReplaceState.

We named this function as withReplaceState because, replaceState is the underline API used for this functionality. Read more about replace state & the history API.

FlowRouter.reload()

FlowRouter routes are idempotent. That means, even if you call FlowRouter.go() to the same URL multiple times, it only activates in the first run. This is also true for directly clicking on paths.

So, if you really need to reload the route, this is the API you want.

FlowRouter.wait() and FlowRouter.initialize()

By default, FlowRouter initializes the routing process in a Meteor.startup() callback. This works for most of the apps. But, some apps have custom initializations and FlowRouter needs to initialize after that.

So, that's where FlowRouter.wait() comes to save you. You need to call it directly inside your JavaScript file. After that, whenever your app is ready call FlowRouter.initialize().

eg:-

// file: app.js
FlowRouter.wait();
WhenEverYourAppIsReady(function() {
  FlowRouter.initialize();
});

For more information visit issue #180.

FlowRouter.onRouteRegister(cb)

This API is specially designed for add-on developers. They can listen for any registered route and add custom functionality to FlowRouter. This works on both server and client alike.

FlowRouter.onRouteRegister(function(route) {
  // do anything with the route object
  console.log(route);
});

Let's say a user defined a route like this:

FlowRouter.route('/blog/:post', {
  name: 'postList',
  triggersEnter: [function() {}],
  subscriptions: function() {},
  action: function() {},
  triggersExit: [function() {}],
  customField: 'customName'
});

Then the route object will be something like this:

{
  pathDef: '/blog/:post',
  name: 'postList',
  options: {customField: 'customName'}
}

So, it's not the internal route object we are using.

Subscription Management

For Subscription Management, we highly suggest you to follow Template/Component level subscriptions. Visit this guide for that.

FlowRouter also has it's own subscription registration mechanism. We will remove this in version 3.0. We don't remove or deprecate it in version 2.x because this is the easiest way to implement FastRender support for your app. In 3.0 we've better support for FastRender with Server Side Rendering.

FlowRouter only deals with registration of subscriptions. It does not wait until subscription becomes ready. This is how to register a subscription.

FlowRouter.route('/blog/:postId', {
    subscriptions: function(params, queryParams) {
        this.register('myPost', Meteor.subscribe('blogPost', params.postId));
    }
});

We can also register global subscriptions like this:

FlowRouter.subscriptions = function() {
  this.register('myCourses', Meteor.subscribe('courses'));
};

All these global subscriptions run on every route. So, pay special attention to names when registering subscriptions.

After you've registered your subscriptions, you can reactively check for the status of those subscriptions like this:

Tracker.autorun(function() {
    console.log("Is myPost ready?:", FlowRouter.subsReady("myPost"));
    console.log("Are all subscriptions ready?:", FlowRouter.subsReady());
});

So, you can use FlowRouter.subsReady inside template helpers to show the loading status and act accordingly.

FlowRouter.subsReady() with a callback

Sometimes, we need to use FlowRouter.subsReady() in places where an autorun is not available. One such example is inside an event handler. For such places, we can use the callback API of FlowRouter.subsReady().

Template.myTemplate.events({
   "click #id": function(){
      FlowRouter.subsReady("myPost", function() {
         // do something
      });
  }
});

Arunoda has discussed more about Subscription Management in FlowRouter in this blog post about FlowRouter and Subscription Management.

He's showing how to build an app like this:

FlowRouter's Subscription Management

Fast Render

FlowRouter has built in support for Fast Render.

  • meteor add meteorhacks:fast-render
  • Put router.js in a shared location. We suggest lib/router.js.

You can exclude Fast Render support by wrapping the subscription registration in an isClient block:

FlowRouter.route('/blog/:postId', {
    subscriptions: function(params, queryParams) {
        // using Fast Render
        this.register('myPost', Meteor.subscribe('blogPost', params.postId));

        // not using Fast Render
        if(Meteor.isClient) {
            this.register('data', Meteor.subscribe('bootstrap-data');
        }
    }
});

Subscription Caching

You can also use Subs Manager for caching subscriptions on the client. We haven't done anything special to make it work. It should work as it works with other routers.

IE9 Support

FlowRouter has IE9 support. But it does not ship the HTML5 history polyfill out of the box. That's because most apps do not require it.

If you need to support IE9, add the HTML5 history polyfill with the following package.

meteor add tomwasd:history-polyfill

Hashbang URLs

To enable hashbang urls like mydomain.com/#!/mypath simple set the hashbang option to true in the initialize function:

// file: app.js
FlowRouter.wait();
WhenEverYourAppIsReady(function() {
  FlowRouter.initialize({hashbang: true});
});

Prefixed paths

In cases you wish to run multiple web application on the same domain name, you’ll probably want to serve your particular meteor application under a sub-path (eg example.com/myapp). In this case simply include the path prefix in the meteor ROOT_URL environment variable and FlowRouter will handle it transparently without any additional configuration.

Add-ons

Router is a base package for an app. Other projects like useraccounts should have support for FlowRouter. Otherwise, it's hard to use FlowRouter in a real project. Now a lot of packages have started to support FlowRouter.

So, you can use your your favorite package with FlowRouter as well. If not, there is an easy process to convert them to FlowRouter.

Add-on API

We have also released a new API to support add-on developers. With that add-on packages can get a notification, when the user created a route in their app.

If you've more ideas for the add-on API, let us know.

Difference with Iron Router

FlowRouter and Iron Router are two different routers. Iron Router tries to be a full featured solution. It tries to do everything including routing, subscriptions, rendering and layout management.

FlowRouter is a minimalistic solution focused on routing with UI performance in mind. It exposes APIs for related functionality.

Let's learn more about the differences:

Rendering

FlowRouter doesn't handle rendering. By decoupling rendering from the router it's possible to use any rendering framework, such as Blaze Layout to render with Blaze's Dynamic Templates. Rendering calls are made in the the route's action. We have a layout manager for React as well.

Subscriptions

With FlowRouter, we highly suggest using template/component layer subscriptions. But, if you need to do routing in the router layer, FlowRouter has subscription registration mechanism. Even with that, FlowRouter never waits for the subscriptions and view layer to do it.

Reactive Content

In Iron Router you can use reactive content inside the router, but any hook or method can re-run in an unpredictable manner. FlowRouter limits reactive data sources to a single run; when it is first called.

We think that's the way to go. Router is just a user action. We can work with reactive content in the rendering layer.

router.current() is evil

Router.current() is evil. Why? Let's look at following example. Imagine we have a route like this in our app:

/apps/:appId/:section

Now let's say, we need to get appId from the URL. Then we will do, something like this in Iron Router.

Templates['foo'].helpers({
    "someData": function() {
        var appId = Router.current().params.appId;
        return doSomething(appId);
    }
});

Let's say we changed :section in the route. Then the above helper also gets rerun. If we add a query param to the URL, it gets rerun. That's because Router.current() looks for changes in the route(or URL). But in any of above cases, appId didn't get changed.

Because of this, a lot parts of our app get re-run and re-rendered. This creates unpredictable rendering behavior in our app.

FlowRouter fixes this issue by providing the Router.getParam() API. See how to use it:

Templates['foo'].helpers({
    "someData": function() {
        var appId = FlowRouter.getParam('appId');
        return doSomething(appId);
    }
});

No data context

FlowRouter does not have a data context. Data context has the same problem as reactive .current(). We believe, it'll possible to get data directly in the template (component) layer.

Built in Fast Render Support

FlowRouter has built in Fast Render support. Just add Fast Render to your app and it'll work. Nothing to change in the router.

For more information check docs.

Server Side Routing

FlowRouter is a client side router and it does not support server side routing at all. But subscriptions run on the server to enable Fast Render support.

Reason behind that

Meteor is not a traditional framework where you can send HTML directly from the server. Meteor needs to send a special set of HTML to the client initially. So, you can't directly send something to the client yourself.

Also, in the server we need look for different things compared with the client. For example:

  • In the server we have to deal with headers.
  • In the server we have to deal with methods like GET, POST, etc.
  • In the server we have Cookies.

So, it's better to use a dedicated server-side router like meteorhacks:picker. It supports connect and express middlewares and has a very easy to use route syntax.

Server Side Rendering

FlowRouter 3.0 will have server side rendering support. We've already started the initial version and check our ssr branch for that.

It's currently very usable and Kadira already using it for https://kadira.io

Better Initial Loading Support

In Meteor, we have to wait until all the JS and other resources send before rendering anything. This is an issue. In 3.0, with the support from Server Side Rendering we are going to fix it.

Migrating into 2.0

Migrating into version 2.0 is easy and you don't need to change any application code since you are already using 2.0 features and the APIs. In 2.0, we've changed names and removed some deprecated APIs.

Here are the steps to migrate your app into 2.0.

Use the New FlowRouter Package

  • Now FlowRouter comes as kadira:flow-router
  • So, remove meteorhacks:flow-router with : meteor remove meteorhacks:flow-router
  • Then, add kadira:flow-router with meteor add kadira:flow-router

Change FlowLayout into BlazeLayout

  • We've also renamed FlowLayout as BlazeLayout.
  • So, remove meteorhacks:flow-layout and add kadira:blaze-layout instead.
  • You need to use BlazeLayout.render() instead of FlowLayout.render()

Stop using deprecated Apis

  • There is no middleware support. Use triggers instead.
  • There is no API called .reactiveCurrent(), use .watchPathChange() instead.
  • Earlier, you can access query params with FlowRouter.current().params.query. But, now you can't do that. Use FlowRouter.current().queryParams instead.