zxbodya/router1

Reactive routing library for universal(isomorphic) web applications

Router1

Reactive routing library for web applications(both server and client side). Built with RxJS.

What makes it different? Why it was created?

• reactive
• declarative
• minimalistic
• feature complete
• ui framework agnostic

It provides:

• url matching
• url generation (by route name and parameters)
• history abstractions for various use cases (server-side, html5 browser, via location.hash, or for tests)
• transition handling (waiting for data, redirect, before enter checks, automatic cancellation of previous transition)
• state updates handling (hashchange, onbeforeunload)
• checking if route is active

Above that it:

• is not coupled to specific UI library, and does not dictate how to structure your view components
• has powerful url expression language, allowing to use regular expressions for url parts
• supports url query parameters
• allows nested routes definition, and is not opinionated about composing handlers for them(it is up to you)
• allows to have conflicting route patterns (and will use the first one with available data)
• allows to pass additional state when navigating
• offers possibility for implementing better ux
  • can wait for minimal required data before transition to new state
  • scrolling top or to anchor only after initial state render(since it waits for minimal required data - it will scroll to the right place on page)
  • not scrolling when navigating backward or forward (browser restores correct scroll position on its own)
  • not finished transition is automatically canceled when new one is started

Installation

Library is available on npm as router1.

When using with React - router1-react would be helpful, it provides all required components.

Also, I tried to make it as un-opinionated as possible, so following parts are not included:

• how rendering should happen
• how scrolling should be done
• when to scroll and when do not

However all this can be quite typical for most of application, and you can see reference implementation here router1-app-template Application template provides webpack build and dev server configurations and routing implementation, also it uses rx-react-container for connecting rxjs logic to react views.

Quick overview

Router consists of following parts

• URL expression language
• history abstraction (for a browser, server, and for testing)
• router itself

Expression language

Url pattern for routes can be written using following constructions:

• /path/name - for static path
• /path/<param> - for path with parameter
• /path/<param:\w+> - for path with parameter that matches regular expression \w+
• /path?q - to add query parameter q to the url

Note: query parameters can be boolean - if parameter is not passed it would be returned as false, if there is no string value - as true

History abstractions

• createServerHistory provides static location from url, meant to be used server side
• createBrowserHistory provides browser location. uses html5 history, when available, or location.assign/replace when history API is not supported
• createHashLocation like createBrowserHistory but using location.hash to store url, usefull for cordova apps or supporting older browsers
• createTestHistory almost the same as createServerHistory but allows to navigate, meant to be used for testing

Routes definition

Routes collection is represented by class RouteCollection, which can be created as new RouteCollection(routes)

Each route is defined as object with following properties:

• name - used to reference specific route (when generating URL, or checking is route active)
• url - URL expression for specific route
• handler or handlers array - "something" that your app will use to handle state associated with the route(more details below)
• routes - array of routes, nested in current one

When declaring nested routes, they are combined, as following:

• name - combination of all name parts separated by dot symbol
• url - combined URL expression
• handlers - array with all of the route handles

When declaring nested routes, all properties are optional and would be skipped.

For example:

const routeCollection = new RouteCollection([
  {
    name: 'home',         
    url: '/home',
    handler: homeHandler,
  },
  {
    name: 'info',         
    url: '/path?query1',
    handler: infoHandler,
    routes: [
      {
        handler: defaultHandler,    
      },
      {
        name: 'article',
        url: '/<articleId:\\d+>',
        handler: contactHandler,
      },
    ],
  },
]);  

Effectively is the same as:

const routeCollection = new RouteCollection([
  {
    name: 'home',         
    url: '/home',
    handlers: [homeHandler],
  },
  {
    name: 'info',         
    url: '/path?query1',
    handlers: [infoHandler, defaultHandler],
  },
  {
    name: 'info.article',         
    url: '/path/<articleId:\\d+>?query1',
    handlers: [infoHandler, contactHandler],
  },
]);  

Handlers in route declaration can be anything you like, that would be enough for state handling.

Router

Router has following options:

• history
• routeCollection
• loadState(transition):Observable loads matched state associated with transition object, if needed data not found and next matching route should be used - should emit false
• renderState(state, transition):Observable - render loaded state, and returns Observable with rendering result
• scrollBehavior object providing onLocationChange and onHashChange methods allowing to implement custom scrolling behavior after state was rendered of location hash changed

loadState is called with transition having following properties:

• route - route definition object
  • name - route name
  • handlers - handlers specified in route configuration
• params - params from matched route
• location - transition location
• forward(url) - method to trigger redirect to specific location
• router - router instance (can be useful to generate redirect or url for example)

In case when no matching route was found route would be set to {name: null, handlers:[]})

• loadState should return observable with object to be latter used in renderState, also to alter navigation behavior it can have following methods:
  • onHashChange({pathname, search, hash, state}) - method would be called when location hash was changed after rendring (not triggered on first render)
  • onBeforeUnload() - callback, that would be called before transition from state or user trying to close the page.
    • should return text message to be displayed in confirm dialogue,
    • or empty string when no confirmation is required

Router can be created as following:

const scrollBehavior= new ScrollBehavior(new ScrollManager());

const router = new Router({
  history: createBrowserHistory(),
  routeCollection,
  loadState,
  renderState,
  // browser scroll behavior; not needed server-side
  scrollBehavior,        
});

Start/stop listening location changes

• start() - subscribe to location changes, and start handling routes (also called when using renderResult)
• stop() - opposite to start, stop listening to location changes

Subscribe to render results:

router.renderResult()
  .forEach(renderResult => {
    // will be called when route was loaded and rendered
    // can be useful for example to track page views
    // or if you are have server side app - you can return rendered HTML here 
  });

To handle onbreforeload browser event there is callback onBeforeUnload in router, it can be used as following:

window.onbeforeunload = router.onBeforeUnload;

Other router public methods are:

• isActive(route, params) - check if route is active
  • if route is prefix of active route - route considered active
  • if some of parameters of active route is not specified it is ignored
• createUrl(name, params = {}, hash = '') - create url for route with params
• navigate(route, params = {}, hash = '', state = {}) - navigate to route with params
• navigateToUrl(url, state = {}) - navigate to specific url